################################################################################ # This is the Changelog for Grid Engine source code changes. # # Possible keywords: "Bugfix", "Belongs to", "Review", "Enhancem.", # # "Merged from", "Cleanup", "Issue", "Bugster" # # # # ===== PLEASE DO NOT DELETE THIS HEADER ==== # ################################################################################ 1 2 3 4 5 6 7 01234567890123456789012345678901234567890123456789012345678901234567890123456789 MD-2009-12-11-0: CVS: Created V62u5_TAG ------------------------------------------------------- LP-2009-12-09-0: Enhancem.: Added missing modified autoinstall template for new -nost option Belongs to: LP-2009-12-07-0 Review: AS MD-2009-12-08-0: Bugfix: Added a missing MoveLog in autoinstall HP-2009-12-07-1: Enhancem.: Improved slotwise preemption description in queue_conf man page Review: TBD LP-2009-12-07-0: Bugfix: Qmaster autoinstallation might never exit if installation template contains invalid SGE_ROOT Bugster: 6907931 Issue: 3205 Enhancem.: Added a header when entering admin_user keystore password Enhancem.: Addedd -nost flag to the inst_sge that allows to disable Sun Service Tags Review: AS, JG SH-2009-12-07-1: Bugfix: xsd-files refered to a wrong URL or URL was not there at all (part 2). Belongs to: SH-2009-12-07-0 Review: JG SH-2009-12-07-0: Bugfix: xsd-files refered to a wrong URL or URL was not there at all (part 1). Bugster: 6885532 Issue: 3145 Review: JG HP-2009-12-07-0: Bugfix: Fixed update bugs in slotwise preemption Review: SH DG-2009-12-02-1: Bugfix: Manual pages qconf(1) - section -sep is not accurate Review: AP DG-2009-12-02-0: Bugfix: execd is crashing because of an uninitialized return value on Solaris when using core binding Review: SH MD-2009-12-02-2: CVS: Created V62u5RC1_TAG ------------------------------------------------------- MD-2009-12-02-1: Enhancem: checkin verison files MD-2009-12-02-0: Bugfix: Fixed broken auto installation error handling. JG-2009-12-01-2: Bugfix: Fixed a typo in qinstance suspend message. CR-2009-12-01-1: Bugfix: qrsh hostname: commlib error on linux - introduced long running shepherds with first fix Belongs to: CR-2009-11-26-1 Belongs to: HP-2009-07-10-0 Bugster: 6899641 Issue: 3190 Review: SH JG-2009-12-01-1: Bugfix: Updated fix for: CPU time in accounting file is wrong for PE jobs with accounting_summary set to TRUE Belongs to: SH-2009-11-23-0 Bugster: 6901597 Issue: 3179 Review: SH CR-2009-12-01-0: Bugfix: qsub -sync does not return correct exit status when qmaster is shutdown and restarted fast Bugster: 6904690 Review: SH EB-2009-12-01-1: Bugfix: Added missing section concerning J2C binding to qsub manpage Review: DG JG-2009-12-01-0: Bugfix: SGE can't handle multi-line environment variables This bug was fixed be GH, I just check it in. Bugster: 6869159 Issue: 3200 Review: JG, EB MD-2009-12-01-0: Bugfix: sgebdb stop does not work Bugster: 6646780 Issue: 3199 Review: SH DG-2009-12-01-0: Cleanup: Missing adoc header added. Review: SH EB-2009-12-01-0: Bugfix: fixed memory leak introduced with EB-2009-11-27 Review: JG HP-2009-11-30-1: Bugfix: o Deleting several simple sleeper jobs with qdel can cause queues going to ERROR or corrupt BDB server This bug was fixed be GH, I just check it in. Bugster: 6894171 Issue: 3198 Review: HP DG-2009-11-30-0: Cleanup: Missing comments (adoc header) added. HP-2009-11-30-0: Bugfix: o configuration of forbidden slotwise preemption relationship structures is not detected Bugster: 6904437 Review: MD JG-2009-11-30-0: Bugfix: fixed memory leak in sge_execd Review: EB EB-2009-11-30-0: Bugfix: fixed memory leak introduced with EB-2009-11-27 Review: JG HP-2009-11-27-0: Bugfix: changing the definition of slotwise preemption relationships doesn't update all previous members Bugster: 6904315 Review: MD CR-2009-11-27-0: Bugfix: upgrade problems with jvm thread when upgradeing from 61 to 62 Bugster: 6885673 Issue: 3139 Review: DG SH-2009-11-27-0: Bugfix: older clients get no error message when trying to connect to qmaster with an higher GDI-version Bugster: 6905290 Issue: 3197 Review: EB EB-2009-11-27-0: Bugfix: o fixed abort of qmaster if global configuration is not available after update. o fixed spooling/parsing problem in qmaster after update. as a result all pending/running jobs of previous GE version were ignored. o fixed problem with CULL namespace if objects should be dumped with lWrite*(). This problem has been introduced with EB-2009-11-20-1 Bugster: 6902905 Review: JG DG-2009-11-26-0: Bugfix: Fixed memory leak when using core binding on Linux. Review: HP CR-2009-11-26-1: Bugfix: qrsh hostname: commlib error on linux Bugster: 6899641 Issue: 3190 Review: JG CR-2009-11-26-0: Bugfix: execd cannot get rid of not existing task which results in qmaster <-> execd protocol loop Bugster: 6858356 Issue: 3170 Review: JG DG-2009-11-25-0: Bugfix: - Fixed memory leaks when using core binding on Solaris. - Fixed wrong processor core allocation on Solaris - Removed duplicate FREE macro Review: SH CR-2009-11-24-0: Bugfix: Interactive jobs get superfluous error message when they can't be scheduled Bugster: 6903978 Issue: 3185 Review: SH HP-2009-11-24-0: Bugfix: o with slotwise preemption, doubly suspended tasks are not always unsuspended correctly Bugster: 6903987 Review: MD MD-2009-11-23-0: Bugfix: o auto installation results in endless loop with wrong SGE_ROOT Bugster: 6903984 Issue: 3191 Review: HP AA-2009-11-23-0: Bugfix: upgrade change for management.properties Review: MD SH-2009-11-23-0: Bugfix: CPU time in accounting file is wrong for PE jobs with accouting_summary set to TRUE Bugster: 6901597 Issue: 3179 Review: EB CR-2009-11-20-0: Bugfix: upgrade problems with max_advance_reservations parameter when upgradeing from 61 to 62 Bugster: 6885690 Issue: 3140 Review: EB DG-2009-11-23-0: Bugfix: reporting of core binding after execd restart contains an additional new line Bugster: 6902868 Review: EB RN-2009-11-20-3: Bugfix: Adding host to hostgroup causes extra slot-only queue instances Bugster: 6903526 Issue: 3184 Review: EB RN-2009-11-20-2: Bugfix: sge_qmaster crashes without a coredump with message from sge_smf_contract_fork() Bugster: 6900836 Issue: 3189 Review: HP EB-2009-11-20-1: Bugfix: o Updating to 6.2u4 causes already running jobs to be "lost Bugster: 6902905 Review: SH DG-2009-11-20-1: Bugfix: upgrade from 6.1 to 62u5 fails because of missing newly introduced complexes Bugster: 6902891 Review: CR DG-2009-11-20-0: Bugfix: wrong qhost help output regarding topology based information switch Bugster: 6901956 Issue: 3183 Review: SH MD-2009-11-19-1: CVS: Created V62u5beta2_TAG ------------------------------------------------------- MD-2009-11-19-0: Bugfix: o fixed broken build CR-2009-11-19-0: Cleanup: o Introduced message macros for INFO loggings Belongs To: RN-2009-11-13-1 Review: RN,MD EB-2009-11-19-0: Bugfix: o core binding is not reported via qstat -j XY -cb after execd shutdown and restart o when using core binding execd messages file reports unsuccessful job deletion o install script used invalid unset command in special error handling case Bugster: 6902501, 6901953 Review: DG DG-2009-11-18-0: Cleanup: Improved qsub man page regarding core binding. Review: EB MD-2009-11-17-1: CVS: Created V62u5beta1_TAG ------------------------------------------------------- HP-2009-11-17-0: Bugfix: Fixed upgrade from previous 6.2u5 to 6.2u5 The new fields for the slotwise preemption are now added automatically if necessary. Review: MD MD-2009-11-17-0: Bugfix: o fstype returns nfs instead of nfsv4 on non-solaris hosts Bugster: 6854020 Issue: 3172 Review: HP DT-2009-11-16-0: Bugfix: o Fixed minor issue with hadoop/setup.pl DG-2009-11-16-0: Bugfix: Removal of new core binding complex reported wrong message. Bugster: 6901517 Review: EB LP-2009-11-16-0: Bugfix: sgemaster script does not print error message when sge_qmaster fails with not zero-exit code Bugster: 6901591 Issue: 3182 Bugfix: sgemaster -migrate does not work if host is using SMF qmaster service Bugster: 6901594 Issue: 3181 Review: HP EB-2009-11-14-0: Bugfix: o JJSV related bugfixed - passing of commandline arguments was incorrect in Java JSV script so that class could not be found. - corrected parsing of ENV JSV command - JJSV can now accept variables without value - added code that handles special characters contained in environment variable values - removed code that automaticaly modifys jobs if some parameters where changed although JJSV did not contain the jsv_correct command - Changed module tests accordingy o Added new Java JSV script that is used in jsv_script TS test Review: TBD DT-2009-11-13-0: Bugfix: o Fixed problems found during additional Herd manual testing o Added hadoop to setfileperm.sh MP-2009-11-13-1: Bugfix: shadowd uninstallation does not remove local host configuration Bugster: 6828316 Issue: 2989 Review: LP MP-2009-11-13-0: Bugfix: fixed spelling mistakes in all man pages DG-2009-11-13-0: Enhancem.: - Added threads in topology string in case of hyper- threading on Linux architectures. - Some clean ups and bug fixes Review: SH, EB RN-2009-11-13-1: Bugfix: qsub -sync does not return correct exit status during qmaster failover Bugster: 6888406 Issue: 3178 Review: CR JG-2009-11-13-0: Bugfix: Fixed memory leaks in test programs and in jsv script Review: CR, EB DT-2009-11-12-0: Bugfix: o Fixed problems found during Herd manual testing - Will rerun tests tomorrow EB-2009-11-11-0: Bugfix: o fixed qstat -j output that caused problem in TS qstat test due to J2C binding changes o changed qrsh related code in shepherd Review: JG (HP) DT-2009-11-10-2: Bugfix: o Added SISSL headers to the remaining properties files DT-2009-11-10-1: Bugfix: o Added SISSL headers to some properties files o Added tests for Herd o Fixed Herd issues found by tests o Corrected error in Herd README HP-2009-11-10-0: Bugfix: Added SISSL headers to HTML and SDA files DT-2009-11-10-0: Bugfix: o Added SISSL headers to build files DG-2009-11-10-0: Bugfix: Fixed hardware thread reporting on Solaris. Review: CR DT-2009-11-09-1: Bugfix: o Added missing hadoop README DG-2009-11-09-0: Enhancem.: Updated showq binary in order to support core binding functionality. Review: SH DT-2009-11-09-0: Bugfix: o Fixed minor issue with Herd build.xml -----------------+------------+------------------------------------------------- MD-2009-11-09-2: CVS: Created V62u5alpha2_TAG ------------------------------------------------------- MD-2009-11-09-1: Bugfix: o Fixed broken hadoop build.xml Enhancem: o added version string MD-2009-11-09-0: Bugfix: o Fixed broken JSV build DT-2009-11-08-1: Bugfix: o Fixed broken JSV build DT-2009-11-08-0: Enhancem.: o Added JavaDocs for Herd o Addressed review comments o Fixed minor issues o Added README for hadoop o Added hadoop.home to build.properties o Moved JSV.jar into SGE_ROOT/lib o Moved DRMAA build to CLASSES/jdrmaa o Moved JSV build to CLASSES/jjsv o Moved Herd build to CLASSES/jjsv o Moved JSV JavaDoc build to JAVADOCS/jjsv o Grouped Java libraries in distinst MD-2009-11-06-0: Bugfix: added translation of hostname from capital letters to small letters to fix the hostcompare. Belongs to: JG-2009-11-06-0 Review: CR DG-2009-11-06-0: Bugfix: loadsensor -loadval reported more than the selected loadvalue Bugster: 6896657 Review: EB JG-2009-11-06-0: Bugfix: SGE distinguish between short host name and full name with the domain Bugster: 6896663 Issue: 3176 Review: MD DG-2009-11-05-0: Bugfix: - Fixed dependencies in Makefile - Fixed execd reporting of bound cores when it finds running jobs while startup - Cleanup and adoc headers Review: CR EB-2009-11-05-0: Bugfix: Added missing man pages for core binding Review: CR HP-2009-11-05-0: Enhancem.: Added section for slotwise sos to man page Changed: queue_conf.5 Review: TBD DG-2009-11-03-1: Bugfix: Invalid free because of invalid mem alloc. Belongs to: JG-2009-10-30-0 Review: SH SH-2009-11-03-0: Bugfix: fixed a segfault caused by an invalid free Belongs to: JG-2009-10-30-0 Review: DG DG-2009-11-03-0: Cleanup: Core binding spec updated. DT-2009-11-02-0: Enhancem.: o Added tests for Herd o Fixed issues uncovered by tests Review: pending EB-2009-11-02-0: Enhancem.: o added support for core-binding in jjsv Review: DT DT-2009-10-30-0: Bugfix: o Added jjsv example to distinst Belongs to: DT-2009-10-23-0 JG-2009-10-30-0: Bugfix: Fixed minor memory leaks. Review: DG DT-2009-10-28-0: Enhancem.: o Added Hadoop integration (Herd) - Requires JDK 1.6 to build - "aimk -only-java herd" to build - Still needs (lots of) documentation - Still needs unit tests Bugster: 6882596 Review: TM, AS JF-2009-10-28-0: Bugfix: allow users to limit concurrent array job task execution jsv and qmon part Belongs to: JF-2009-08-24-0 Bugster: 6684887 Issue: 2542 Review: EB, AA HP-2009-10-28-1: Bugfix: - Fixed lib path issue on win32-x86 - Added more detailed debug output to IJS Review: JG HP-2009-10-28-0: Enhancem.: Slotwise suspend on subordinate Bugster: 6538365 Issue: 35 Review: JG, EB DG-2009-10-28-1: Cleanup: Enabled ENABLE_BINDING execd param in order to to turn core binding feature on per host. Review: SH EB-2009-10-28-1: Bugfix: fixed JSV-core-binding-related bugs Review: DG SH-2009-10-29-0: Bugfix: libraries opened with dlopen and the RTLD_DEEPBIND flag does not use our jemalloc mechanism. This causes a segfault. Bugster: 6840703 Issue: 3193 Review: HP DT-2009-10-28-0: Bugfix: o Corrected problem with jjsv build properties Belongs to: DT-2009-10-23-0 EB-2009-10-28-0: Bugfix: fixed JSV-core-binding-related bugs Review: DG DG-2009-10-28-0: Bugfix: Reporting of binding settings from execd to client. Review: EB HP-2009-10-27-0: Bugfix: Fixed broken build on win32-x86 Changed: utilbin/Makefile, utilbin/loadcheck.c AA-2009-10-27-0: Bugfix: - sgeinspect for windows cannot connect in SSL mode - management.properties remove keystorePassword from management.properties - minor test fixes - JGDIAgent state fix Bugster: 6849311,6876508 Review: RH, AF LP-2009-10-27-2: Bugfix: JMX part of qmaster installation should verify JVM is 32-bit or 64-bit Bugster: 6876446 Issue: 3120 Review: SH DT-2009-10-27-1: Bugfix: o Backported jsv classes to 1.5 Belongs to: DT-2009-10-23-0 LP-2009-10-27-1: Enhancem.: Improved messages in sgemaster script Belongs to: LP-2009-10-20-0 Bugfix: Upgrade load_sge_config.sh now suggests builtin value for IJS on error Review: SH DT-2009-10-27-0: Bugfix: o Backported libs/jjsv/build.xml to Ant 1.6.5 o Fixed minor jjsv javadoc error o Added jjsv wrapper script Belongs to: DT-2009-10-23-0 LP-2009-10-27-0: Bugfix: Windows execd installation does not create proper certificates Bugster: 6895510 Issue: 3168 Review: JF CR-2009-10-27-0: Bugfix: possible race condition in drmaa implementation when event clients report errors Bugster: 6895073 Issue: 3167 Merged from: V61_BRANCH Review: DG EB-2009-10-27-0: Enhancem.: added jsv support for j2c binding parameters Review: DG DG-2009-10-27-0: Ehnhacem.: Enhanced loadcheck binary in order to print out processor topology related information. Review: EB, SH DT-2009-10-26-0: Enhancem.: o Added JSV Java lang binding into aimk o Backported JSV Java lang binding tests onto junit 3.8.2 - Still missing testsuite integration Belongs to: DT-2009-10-23-0 AA-2009-10-26-0: Enhancem.: additional permissions for new keystore.password file and modifyThread problem Review: RH JG-2009-10-26-0: Bugfix: merged minor memory leaks from V61_BRANCH, cleanup. Review: CR DT-2009-10-23-0: Enhancem.: o Initial commit of JSV Java language binding - Not yet added into aimk or the testsuite Bugster: 6882592 Review: EB DG-2009-10-23-0: Enhancem.: Added -binding pe functionality for pe_hostfile. Review: CR MD-2009-10-23-1: Enhancem.: o Added missing tarball MD-2009-10-23-0: Enhancem.: o Added 6.1u7 gdi version entry to version.c o Added, updated webfiles for 6.2u4 courtesy binary download DG-2009-10-22-1: Enhancem.: Added help output for binding parameter in qsub, qrsh, qsh and qalter. Review: SH DG-2009-10-22-0: Enhancem.: Core binding option 'env'. Minor cleanups. Review: SH ZG-2009-10-22-0: Bugfix: gui-installer: fixed automated installation support Review: LP MP-2009-10-21-0: Bugfix: jsv.pm does not recognize -A option when used with qsub Bugster: 6883285 Issue: 3151 Review: TB DG-2009-10-21-1: Cleanup: - consistent use of binding_type - memory leak Review: EB SH-2009-10-21-0: Bufix: Make aimk -no-plpa work again Belongs to: DG-2009-10-21-0 Review: DG DG-2009-10-21-0: Enhancem.: - PLPA is now a static lib - Algorithm for linear binding added - Cleanups EB DG-2009-10-20-1: Cleanup: Broken build on darwin fixed. Review: EB DG-2009-10-20-0: Enhancem.: Job2Core binding spec added. Review: EB LP-2009-10-20-1: Bugfix: Shadowd might be started on host which hostname is a prefix on real shadow host Bugster: 6892908 Issue: 3164 Belongs to: LP-2009-10-20-0 Review: ZG LP-2009-10-20-0: Bugfix: Shadowd startup scripts do not work Bugster: 6892804 Issue: 3163 Bugfix: Shadowd might be started on host which hostname is a prefix on real shadow host Bugster: 6892908 Issue: 3164 Bugfix: sgemaster script might return 0 on failure Belongs to: LP-2009-04-09-0 Bugster: 6827877 Issue: 2987 Review: MD EB-2009-10-20-1: Enhancem.: Job2Core binding: fixed bug in qmon Review: DG EB-2009-10-20-0: Enhancem.: Job2Core binding: client related things and infrastructure Review: DG DG-2009-10-15-0: Cleanup: Fixed core binding parsing code. Added more output on execd messages. Fixed generation of OS internal processor ids. Review: SH DG-2009-10-14-0: Cleanup: Job2Corebinding related cleanups und bug fixes. Review: EB LP-2009-10-14-0: Enhamcem.: Improved/clarified gui-installer host panel help page about resolvable state Review: JO DG-2009-10-13-0: Cleanup: Missing copy of selected binding strategy in parsing command line function added. Review: EB DG-2009-10-12-2: Cleanup: - aimk reduces environment for plpa lib configure - added function for parsing command line everything for Job2Core binding Review: EB SH-2009-10-12-0: Bugfix: Add ability to commlib to wait for data on external file descriptors Bugster: 6860857 Review: CR DG-2009-10-12-1: Enhancem.: Moved functions from obj to uti. Smaller fixes and functionality added. (Job2Core binding). Review: EB DG-2009-10-12-0: Cleanup: Avoid casting from const (compiler warning). Review: CR JF-2009-10-08-0: Bugfix: allow users to limit concurrent array job task execution qalter part Belongs to: JF-2009-08-24-0 Bugster: 6684887 Issue: 2542 Review: EB DG-2009-10-08-0: Bugfix: Fixed segfault when kstat cpu_info returns nothing. Review: SH JG-2009-10-06-0: Bugfix: qmaster aborts when submitting a job depending on a just finishing tightly integrated parallel job Bugster: 6883957 Issue: 3152 Review: EB DG-2009-10-05-0: Cleanup: Fixed broken build on ULINUXAMD64_24 Review: SH JG-2009-10-05-0: Bugfix: modifying the global configuration triggers spooling of all exec hosts Bugster: 6880212 Issue: 3129 Review: CR DG-2009-10-04-0: Cleanup: Fixed broken build on ULINUX86_24 Review: SH DG-2009-10-03-0: Enhancem.: Initial dirty checkin of core binding feature for collaboration purposes. Do not use any new parameters. Review: EB SH-2009-10-02-0: Enhancem.: Added -no-mtmalloc option in aimk to disable linking against the mtmalloc-library. This is needed to use memory leak detection in dbx. Review: DG AP-2009-10-01-0: Bugfix: Fixed product version comparison in java drmaa test. Review: DG DG-2009-10-01-0: Cleanup: Make PLPA header accessible to daemons. Review: AA DG-2009-09-29-0: Cleanup: Do not inherit env to PLPA configure script. Review: AA DG-2009-09-28-0: Enhancem.: Added PLPA library and installation and build support for core binding in Linux. Review: AA MD-2009-09-24-0: CVS: Created V62u4_TAG ----------------------------------------------------- MD-2009-09-23-2: Enhancem.: raised GDI version MD-2009-09-23-1: Bugfix: fixed broken build MD-2009-09-23-0: Cleanup: fixed issues reported by insure, forgot to check in. Belongs to: MD-2009-09-21-0 Review: CR JG-2009-09-22-0: Bugfix: fixed a memory leak in qmaster Review: EB, MD MD-2009-09-22-0: Enhancem.: check in file containig version string 6.2u4 CR-2009-09-21-0: Cleanup: commlib did check for timed-out messages in application messages queue every second. Since the timeout is 3600 seconds the check now is done once per minute. Review: MD MD-2009-09-21-0: Bugfix: fixed issues reported by insure Review: CR HP-2009-09-14-0: Bugfix: Fixed broken build on most systems because of checkin RN-2009-09-04 MD-2009-09-12-0: Bugfix: qmaster did not start, resulte of a leak fix. Belongs To: MD-2009-09-11-0 MD-2009-09-11-0: Bugfix: fixed memory leaks found with insure Review: EB SH-2009-09-11-0: Bugfix: With kernel 2.6 in Linux there is no timestamping on the procfs anymore. Therefore the suggested fix does not work. Bugster: 6749299 Issue: 2962 Review: JG RH-2009-09-11-0: Bugfix: jgdi can not convert exec hosts objects with reporting variable (temp fix that does not require an update procuedure for 6.2u4 release) Bugster: 6880144 Issue: 3128 Review: AA LP-2009-09-10-0: Bugfix: Automatic execd uninstallation does not work on CentOS r4u5 Bugster: 6880888 Issue: 3130 Bugfix: Automatic execd uninstallation might not work Bugster: 6880894 Issue: 3131 Review: JG MP-2009-09-10-0: Bugfix: made modification of checkin from MP-2009-08-20-0 Bugster: 6865717 Issue: 3111 Review: EB RN-2009-09-04-0: Bugfix: Scheduler crashes with got NULL element for CE_requestable Bugster: 6873408 Issue: 3126 Review: GH RN-2009-09-04-1: Bugfix: sge clients/daemons segfault when bootstrap file is corrupted Bugster: 6876458 Issue: 3127 Review: JG JG-2009-09-05-1: Bugfix: Fixed memory access problem in qstat -xml output. Belongs to: EB-2009-08-05-0 Review: MD JG-2009-09-05-0: Enhancem.: mark new qsub -tc switch as experimental in -help output Belongs to: JF-2009-08-24-0 Review: MD LP-2009-09-03-0: Bugfix: management.properties file should only be readable by admin user or root Bugster: 6876508 Issue: 3123 Belongs to: LP-2009-09-01-0 Review: MD MP-2009-09-03-0: Bugfix: Added missing man page information to submit man page for check in MP-2009-08-25-0 Bugster: 6865717 Issue: 3111 Review: AS LP-2009-09-01-0: Bugfix.: CSP qmaster installation displays confusing message that suggests a failure if qmaster=execd Bugster: 6875947 Issue: 3117 Bugfix: management.properties file should only be readable by admin user or root Bugster: 6876508 Issue: 3123 Bugfix: SMF services should be started even when different SMF service starts them Bugster: 6877588 Issue: 3124 Bugfix: Upgrade JMX logic misleading Bugster: 6877580 Issue: 3125 Review: MD GH-2009-08-31-0: Bugfix: too many qsub -sync jobs wedge the qmaster on failover Bugster: 6873140 Issue: 3113 Review: JG DG-2009-08-31-0: Bugfix: processor set does not work on Solaris architecture Bugster: 6861758 Issue: 3122 Review: JG MP-2009-08-25-0: Bugfix: fixes to initial checkin MP-2009-08-20-0 Bugster: 6865717 Issue: 3111 Review: HP, EB JF-2009-08-24-0: Bugfix: allow users to limit concurrent array job task execution Bugster: 6684887 Issue: 2542 Review: HP MP-2009-08-20-0: Bugfix: jsv Protocol is not handled correctly in client JSV Bugster: 6862641 Issue: 3089 Bugfix: need means to set the jsv command timeout Bugster: 6865717 Issue: 3111 Review: HP RN-2009-08-21-0: Bugfix: Original primary group vanishes after newgrp command (USE_QSUB_GID=true) Issue: 2342 Bugster: 6590010 Review: JG Bugfix: interactive jobs (qlogin, qrsh without command) don't set the TZ environment variable correctly Issue: 2716 Bugster: 6744724 Review: HP LP-2009-08-20-0: Bugfix.: SGE 6.2 auto installation does unnecessary check on SGE_JMX_PORT and prints a misleading suggestion Bugster: 6831541 Issue: 2724 Bugfix: RC script removal is broken if SMF is used Bugster: 6760650 Issue: 3104 Bugfix: Berkeley DB server auto installation does not work when HOSTNAME_RESOLVING=true Bugster: 6806917 Issue: 2944 Bugfix: permission problem during local config creation Bugster: 6830974 Issue: 3105 Bugfix: should ensure qmaster spool dir is empty before continuing the qmaster installation Bugster: 6831541 Issue: 3106 Bugfix: smf support was not enabled but getting a error with SMF dependecy error Bugster: 6832008 Issue: 3107 Bugfix: Upgrade script from 61 to 62 requires working ssh connection to remote hosts Bugster: 6848042 Issue: 3108 Bugfix: Calling inst_sge -auto with non-existing file argument leads to an error Bugster: 6848095 Issue: 3109 Bugfix: inst_sge -copycerts no longer works when no JMX or CSP mode used Bugster: 6856564 Issue: 3110 Bugfix: JMX gets disabled in the autoinstallation if SGE_JVM_LIB_PATH set is to "" Bugster: 6874070 Issue: 3102 Review: AP RM-2009-08-17-0: Bugfix.: Resource leak in juti.jar Bugster: 6869559 Issue: 3098 Review: LP LP-2009-08-17-1: Bugfix.: Qmaster installation might fail when more than 1 qmaster is installed in parallel on the same host Bugster: 6872473 Belongs to: LP-2009-08-17-0 Review: AP LP-2009-08-17-0: Bugfix.: Qmaster installation might fail when more than 1 qmaster is installed in parallel on the same host Bugster: 6872473 Review: AP EB-2009-08-14-0: Enhancem.: added script that might be used to start a bunch of jobs as qmake job. As input it expects a task or makefile. Review: TBD EB-2009-08-05-0: Enhancem.: optimized lGetPosInDescr Review: (RD partially), DG CR-2009-08-05-1: Bugfix: qsub -sync does not connect to shadow master after failover Bugster: 6867897 Issue: 3096 Review: JG CR-2009-08-05-0: Bugfix: qmon does not reconnect to qmaster after shadowd takeover Bugster: 6868133 Issue: 3097 Review: JG SH-2009-08-04-0: Bugfix: online usage of maxvmem incorrect for tightly integrated parallel jobs Bugster: 6287864 Issue: 1314 Review: JG JG-2008-08-03-0: Bugfix: ACCT_RESERVED_USAGE is broken Bugster: 6867805 Issue: 3094 Review: HP MD-2008-07-31-0: Bugfix: o inst_sge -ux does not report error if hostname resovling is inconsistent Bugster: 6849718 Issue: 3043 Review: RH JG-2008-07-30-0: Bugfix: more spooling ops than needed during job start order processing for seq jobs Bugster: 6773668 Issue: 2804 Review: EB MP-2009-07-29-0: Bugfix: qmaster uninstallation fails when BDB RPC server is used Bugster: 6688019 Issue: 3064 Review: JG MD-2008-07-28-0: Bugfix: Submit client or master might fail if JSV is enabled and environment contains fprint format spec Bugster: 6862580 Issue: 3088 Review: EB MD-2009-07-23-0: Cleanup: removing unused INCLUDES from source code (part 1) Issue: 3090 JG-2009-07-22-1: Cleanup: Added documentation for CULL field attributes. Fixed some ADOC headers. Review: EB JG-2009-07-22-0: Bugfix: qstat -help exits 1 Bugster: 6862617 Issue: 3091 Review: EB JG-2009-07-17-0: Bugfix: task spooling always causes job be spooled as well for array jobs with classic spooling Bugster: 6774120 Issue: 2805 Review: EB MP-2009-07-16-0: Bugfix: syntax error for empty SGE_QMASTER/EXECD_PORT Bugster: 6750689 Issue: 3076 Review: JG CR-2009-07-13-0: Cleanup: Added some shepherd trace file logging for INTERIX and TODO statements Review: DG GH-2009-07-12-0: Bugfix: Fixed missing xyz_is_consumable_job parameters in job config file Belongs to: GH-2009-07-08-0 HP-2009-07-10-0: Cleanup: Cleanup of builtin IJS Cleanups without CR: o improved error handling in 20-30 places - output of error messages from lib functions, added missing error handling, ... o code cleanup - indention, brackets, variable names, comments and ADOC headers o got rid of some global and some module variables o improved command parsing in qsh.c - now a list instead of a static array is used to collect the tokens, an allocated array is the generated from the list o if this command line contains quoted arguments, they are parsed as one token now o moved large code blocks to separate functions Bugster: 6857639 Issue: 3075 Review: EB GH-2009-07-10-0: Bugfix: SGE does not handle multiline variable using qsub -v Bugster: 6855413 Issue: 3085 Review: CR DG-2009-07-10-0: Bugfix: Fixed broken installation. Belongs to: GH-2009-07-09-0 Review: CR CR-2009-07-10-0: Cleanup: Fixed possible free(NULL) problem in commlib logging in the case when running out of memory. Merged from: V61_BRANCH (CR-2009-07-10-0) Review: SH GH-2009-07-09-0: Bugfix: execd installation on Windows doesn't provide global spool dir as default, even if it's local Bugster: 6802751 Issue: 3081 Review: HP, MD Bugfix: windows exec autoinstall fails if spooldir was already created with wrong perms/ownership Bugster: 6822285 Issue: 3082 Review: HP, MD Bugfix: autoinstall of Windows exec host hangs in GetLocalSpoolDir Bugster: 6822285 Issue: 3083 Review: HP, MD Bugfix: Windows exechost install failed to create the local spooldir Bugster: 6798623 Issue: 3084 Review: HP, MD JF-2009-07-08-0: Bugfix: Man page should mention reprioritize_interval is coupled to scheduler_interval Bugster: 6786258 Issue: 3078 Review: MP Bugfix: Relationship between suspend_threshold and scheduler_interval needs to be documented Bugster: 6291037 Issue: 1636 Review: MP Bugfix: Man page entry for SCHEDULER_TIMEOUT incorrectly states that default is 600 Bugster: 6806918 Issue: 3079 Review: MP GH-2009-07-08-0: Bugfix: h_vmem setting as JOB consumable fails to properly set the virtual memory limit of a job Bugster: 6823365 Issue: 3068 Review: JG, MP EB-2009-07-06-1: Cleanup: o replaced list implementation in master and schedd by utilib implementation Review: MD EB-2009-07-06-0: Enhancem.: o prepared list names for code generation Review: MD MP-2009-07-02-0: Bugfix: automatic uninstallation failed when using -jmx option Bugster: 6747579 Issue: 2727 Review: CR DG-2009-07-02-0: Cleanup: wrong size when comparing strings Review: JG JG-2009-07-02-0: Cleanup: Removed outdated spooling to postgres database. Review: DG DG-2009-07-01-0: Bugfix: qmod -usj jid returns error status even though job is suspended Bugster: 6842457 Issue: 3071 Review: CR SH-2009-07-01-0: Cleanup: fixed broken build on ulx24-x86 Belongs to: SH-2009-06-24-0 Review: DG MP-2009-06-30-0: Bugfix: cluster name check is failing in installation Bugster: 6659197 Issue: 3046 Review: LP, MD MP-2009-06-29-0: Bugfix: autoinstall fails if spooling dir (BDB) already exits without any error mgs/log file being created Bugster: 6618009 Issue: 3063 Review: JG RN-2009-06-26-0: Bugfix: qalter -w p reports running jobs for jobs that are rescheduled due to queue error Bugster: 6847548 Issue: 3040 Review: JG MP-2009-06-25-0: Bugfix: inadequate inst_qmaster/inst_execd diagnosis if binary and common packages are not in $SGE_ROOT Merged from MP-2008-12-01-0 in V61_BRANCH Bugster: 6617928 Issue: 2807 Review: JG MD-2009-06-25-0: Bugfix: inst_sge shows wrong dialogs during csp certificate installation Bugster: 6834894 Issue: 3066 Review: CR EB-2009-06-24-1: Cleanup.: renamed some sl functions Review: HP, CR EB-2009-06-24-0: Enhancem.: implemented thread safe list that will replace different list implementations in GE code. Review: JG, MD SH-2009-06-24-0: Bugfix: the execution daemon on Linux generate huge CPU load based on monitoring processes Bugster: 6749299 Issue: 2962 Enhancem.: build pdc as standalone binary Review: JG MD-2009-06-23-0: Enhancem.: o updated webpages on sunsource.net o added announcement page for 62u3 o updated 62patches.txt file JG-2009-06-22-0: Bugfix: bad RQS syntax can crash/hang qmaster process Bugster: 6810411 Issue: 2899 Review: DG MD-2009-06-18-0: Bugfix: inst_sge -copycerts exits without error message when called for a cluster having no certificates Bugster: 6769515 Issue: 3062 Review: SH EB-2009-06-17-0: Cleanup: separated CULL object definition from other C code removed superfluous includes corrected some includes eleminated link dependency to sgeobj if some cull function was used Review: JG MD-2009-06-17-0: Bugfix: $SGE_ROOT/util/setfileperm.sh does not change ownership of $SGE_ROOT/start_gui_installer Bugster: 6808082 Review: HP DG-2009-06-17-2: Bugfix: qsub man page does not mention the existance of $JOB_SCRIPT environment variable within a job Bugster: 6851538 Issue: 3058 Bugfix: qalter man page does not imply that qalter works also on running jobs Bugster: 6849741 Issue: 3059 Review: JG DG-2009-06-17-1: Bugfix: DrmaaSuite.java imports are too broad Bugster: 6808646 Issue: 2924 Review: SH ZG-2009-06-17-0: Update: update html man pages Review: LP DG-2009-06-17-0: Bugfix: man page for drmaa_control() refers to deprecated parameters Bugster: 6713724 Issue: 3056 Review: SH JG-2009-06-17-1: Bugfix: fixed a minor memory leak in qstat client Review: EB JG-2009-06-17-0: Bugfix: removing admin user from the manager list leads to a qmaster dead lock Bugster: 6684465 Issue: 3053 Review: EB JG-2009-06-16-0: Bugfix: man page of sge_ckpt points to wrong reference in queue_conf Cleanup: Some code cleanup (formatting). Bugster: 6494492 Issue: 3055 Review: DG MP-2009-06-15-0: Bugfix: qconf -ke[j] must return non-zero exit code on failure Merged from V61_BRANCH MP-2009-06-04-0 Bugster: 6675570 Issue: 2942 Review: HP MD-2009-06-15-1: CVS: Created V62u3_TAG ----------------------------------------------------- MD-2009-06-15-0: Enhancem.: changed version strings to 6.2u3 JG-2009-06-03-0: Bugfix: Fix for memory leak fix JG-2009-05-28-0. Fixed a potential qconf client core dump. Belongs to: JG-2009-05-28-0 Review: EB CR-2009-06-02-1: Bugfix: merged requested changes from LP Review: SH CR-2009-06-02-0: Revert: Reverted changes. Fix seems to have to much effect on current code base and will go into next release. Belongs to: SH-2009-05-29-0 Review: SH JG-2009-06-02-0: Bugfix: qconf -sstree output wrong error message, if there was no sharetree. Belongs to: RD-2009-05-04-0 Bugster: 6847167 Review: CR CR-2009-05-29-1: Bugfix: Fixed typo in path for gethostname save_sge_config.sh Belongs to: CR-2009-05-29-0 Review: DG CR-2009-05-29-0: Bugfix: Hotfix: wrong source command for sh scripts Belongs to: SH-2009-05-29-0 Review: DG SH-2009-05-29-0: Bugfix: Let upgrade scripts use ResolveHosts function Bugster: 6845858 Issue: 3038 Review: CR DG-2009-05-29-0: Bugfix: Installation problems when using non default cellname. Hostcompare does not work correctly. Review: CR MP-2009-05-28-1: Bugfix: spelling errors in the man pages Bugster: 6845162 Issue: 3041 MP-2009-05-28-0: Bugfix: qhost man page needs to clarify how it handles multiple resource attribute requests Bugster: 6843269 Issue: 3039 Review: AS RN-2009-05-28-0: Bugfix: qalter -w p output fails to match qstat -j output Bugster: 6833511 Issue: 3040 Review: RD DG-2009-05-28-0: Cleanup: Autoinstall template comments for JMX port updated. Review: CR CR-2009-05-28-0: Bugfix: o fixed typo in inst_sge which disables auto installation with enabled jmx thread o improved jvm lib path checking in inst_common.sh Review: ZG JG-2009-05-28-1: Enhancem.: Scheduler optimization with many complex variables. Review: AH, EB JG-2009-05-28-0: Bugfix: Fixed minor memory leaks (mostly introduced in 6.2u3 development). Review: EB CR-2009-05-26-0: Cleanup: Fixed copyright headers in source/dist/mpi dir and source/dist and source/clients/gui-installer and source/daemons/shepherd Review: DG LP-2009-05-22-2: Enhancem.: Removed -no-jmx install flag Enhancem.: Wrong FS type message not displayed during auto installation Review: ZG, HP LP-2009-05-22-1: Bugfix: showq does not compile on OpenSuse 11.1 Review: RD LP-2009-05-22-0: Bugfix: gui-installer: CSP installation might fail when qmaster!=startup node Bugster: 6831094 Belongs to: ZG-2009-05-06-0 Bugfix: gui-installer: Exception when reading empty line in managemenent.properties file Bugster: 6844300 Bugfix: gui-installer: Empty JMX screen displayed in custom standalone shadowd installation Bugster: 6844305 Review: ZG MD-2009-05-22-0: Bugfix: english language issues in install_qmaster Bugster: 6810076 Issue: 3037 Review: JG ZG-2009-05-21-0: Bugfix: gui-installer: the "Please wait..." text on ActionPanel is invisible during long last actions in the background. Review: LP JG-2009-05-20-0: Bugfix: accounting_summary in PE only adding cpu/mem/io Bugster: 6774196 Issue: 2787 Review: AP CR-2009-05-19-1: Cleanup: fixed some copyright headers Review: AA CR-2009-05-19-0: Cleanup: fixed some copyright headers Review: TB RD-2009-05-19-0: Bugfix: corrected error message for EXCL complex without consumable flag Bugster: 6842644 Issue: 3034 Review: JG HP-2009-05-15-0: Bugfix: redirecting "qrsh -pty yes" output to a file or a pipe doesn't work Changed: clients/common/sge_client_ijs.c Bugster: 6841529 Issue: 3031 Review: MD RH-2009-05-15-0: Enhancem.: startup script for jgdi eventmon improved Review: AA DG-2009-05-15-0: Bugfix: wrong argument syntax could cause qmaster crash Bugster: 6848366 Review: JG HP-2009-05-14-0: Enhancem.: Improved error messages Changed: common/sge_ijs_comm.c RD-2009-05-13-0: Enhancem.: further qacct optimizations from Ron Chen and Rayson Ho Belongs to: RD-2009-05-07-0 Review: JG, CR JF-2009-05-13-0: Bugfix: qacct does not resolve complex shortcut names Bugster: 6499573 Issue: 3024 Review: JG LP-2009-05-12-0: Updated log description Belongs to: ZG-2009-05-06-0 DG-2009-05-12-1: Cleanup: Fixed typo in man page. Review: CR DG-2009-05-12-0: Cleanup: Fixed typo in man page. Review: CR LP-2009-05-11-0: Bugfix: auto_install does not set the additional_jvm_args which limits qmaster size Bugster: 6827662 Issue: 3007 Belongs to: LP-2009-04-14-0 Bugfix: gui-installer: qmaster installation validation always reports JMV lib missing if not set properly Bugster: 6835592 Issue: 3027 Enhancem.: Detecting JAVA message displaye only when really doing the detection Review: ZG, MD RN-2009-05-11-0: Bugfix: qhost failed to filter the resource request correctly Bugster: 6782283 Issue: 3026 Review: JG JG-2009-05-11-0: Bugfix: inconsistent signal delivery behavior with and without -notify Bugster: 6811202 Issue: 3017 Review: HP JF-2009-05-11-0: Bugfix: man qmod synopsis mentions only queue and omits job Bugster: 6838220 Issue: 3023 Review: AP Bugfix: exechost wrongly named "exec server" in qconf -help Bugster: 6838070 Issue: 3020 Review: AP AP-2009-05-11-0: Bugfix: the path to PE_HOSTFILE variable may not be set for pe slave tasks Bugster: 6825385 Issue: 3018 Review: JG DG-2009-05-11-0: Bugfix: SGE util/arch script is broken for AIX 5.3 Operating System Belongs to: MP-2008-05-15-0 Bugster: 6648160 Issue: 2572 Review: JG CR-2009-05-08-0: Bugifx: qmaster fail-over results in very slow execd reconnect Bugster: 6823756 Issue: 2900 Review: SH, RD RD-2009-05-07-1: Bugfix: fixed broken build Belongs to: RD-2009-05-07-0 RD-2009-05-07-0: Bugfix: improved qacct performance Bugster: 6838487 Issue: 2510 Review: CR ZG-2009-05-06-0: Enhancem.: gui-installer: improved JVM library validation in check_host script Bugfix: gui-installer: admin/submit error handling is broken Bugster: 6806241 Bugfix: gui-installer: CSP installation might fail when qmaster!=startup node Bugster: 6831094 Review: LP ZG-2009-05-05-1: Reverting: gui-installer: using gethostbyname/gethostbyaddr binaries instead of InetAddress class ZG-2009-05-05-0: Docs: gui-installer: Update help pages Enhancem.: gui-installer: Show "Please wait..." text in case of longer procedures on action panels Enhancem.: gui-installer: using gethostbyname/gethostbyaddr binaries instead of InetAddress class Enhancem.: gui-installer: fix old and implement new JUnit tests Bugfix: gui-installer: JVM library path is not mandatory on Configuration window. Bugfix: gui-installer: JMX port has to have the value of SGE_QMASTER_PORT + 2 by default Bugfix: gui-installer: when moving columns the sorting stays in the column index not on the column ifself Bugster: 6786350 Bugfix: gui-installer: let's the user be informed about the user who is running the installer Bugster: 6831971 Review: LP DG-2009-05-04-0: Cleanup: Upgrade routine is misleading when asking for existing backup directory from configuration of the old cluster. Review: JG, TB RD-2009-05-04-0: Bugfix: fixed IDs of classic spooled sharetree Bugster: 6836897 Issue: 3016 Review: JG JG-2009-05-04-0: Bugfix: qmaster crash in schedd_mes_add_global/lGetPosViaElem Bugster: 6827155 Issue: 3014 Review: RD DG-2009-04-30-0: Bugfix: parallel job submission verfication result differs when turning on non-restricting resource quota Bugster: 6832457 Issue: 3013 Review: RD JG-2009-04-29-1: Bugfix: fixed missing depencency for libspoolc.so Review: RD JG-2009-04-29-0: Bugfix: problem with broken dot-file writing for sequential jobs in classic spooling mode Bugster: 6773642 Issue: 2801 Review: EB LP-2009-04-28-1: Updated log description Belongs to: LP-2009-04-03-0 CR-2009-04-28-0: Bugfix: fixed JSV for qalter "-N" option which turned out not to work because of fix for CR 6824346 (qalter using job name as job id was broken) Belongs To: MD-2009-04-24-0 Cleanup: improved some execd message loggings Review: EB, RD LP-2009-04-28-0: Bugfix: Install script do not recognize OpenJDK as valid Java version Bugster: 6834901 Issue: 3008 Review: AP RH-2009-04-28-0: Bugfix: jgdi eventmon did not use credentials from keystore Review: AA RD-2009-04-27-0: Bugfix: fixed write lock starvation Bugster: 6834507 Issue: 3010 Review: EB, AA RN-2009-04-27-0: Bugfix: loadsensor infrastructure in the ibm platform is inconsistent Bugster: 6816902 Issue: 3009 Review: JG JG-2009-04-24-0: Bugfix: make utilbin//rsh ignore SIGUSR1/2 Belongs to: JG-2009-04-23-0 Review: HP MD-2009-04-24-0: Bugfix: qalter -h U fails with CULL error Bugster: 6824346 Issue: 2972 Review: CR AA-2009-04-23-0: Bugfix: fixed authuser root check to allow test install - cause of login failure sgeinspect in test installation mode Bugster: 6822025 Review: RD HP-2009-04-23-0: Bugfix: Fixed broken build on win32-x86 and aix43 in consequence of checkin RD-2009-04-22-0 Changed: libs/sgeobj/str2nm_converter.c/.h JG-2009-04-23-0: Bugfix: we need the ability for qrsh to ignore SIGUSR1 and SIGUSR2 to support -notify Bugster: 6802351 Issue: 2986 Review: HP RD-2009-04-23-0: Bugfix: qconf -Mc returned always 0 even in case of error Belongs to: RD-2009-03-24-0 Review: CR AA-2009-04-22-0: Bugfix: jgdi SSL connections from one client jvm to different SGE cluster might not work Bugster: 6831573 Issue: 3005 Review: RH RD-2009-04-22-0: Added showq binary (Contribution by TACC) Review: DG HP-2009-04-22-0: Fixed broken build on Samba shares Changed: libs/japi/Makefile Review: HP LP-2009-04-22-0: Updated log description Belongs to: LP-2009-04-14-0 AP-2009-04-21-0: Bugfix: qconf -dhgrp can delete a host group referenced from a queue instance Bugster: 6619384 Issue: 2408 Review: DG CR-2009-04-21-0: Bugfix: shadowd startup script returns error if SMF support is enabled Bugster: 6832406 Issue: 3006 Review: LP MD-2009-04-21-0: CVS: Created V62u3beta_TAG ----------------------------------------------------- LP-2009-04-17-1: Bugfix: CheckRunningDeamon for qmaster returns true even if no pid file created Belongs to: LP-2009-04-17-0 Review: CR LP-2009-04-17-0: Bugfix: CheckRunningDeamon for qmaster returns true even if no pid file created Bugster: 6824148 Issue: 2969 Review: CR RC-2009-04-17-0: Enhancem.: better handling of SGE_ARCH_STRING for FreeBSD and NetBSD (from Rayson, Kevin) Review: HP, BD, RC MP-2009-04-16-0: Bugfix: Changing CMDARG[1-n] always changes the first commandline argument in the jsv script. Also the number of arguments is handled as read-only parameter Bugster: 6825381, 6825376 Issue: 2998, 2999 Review: EB LP-2009-04-16-2: Bugfix: gui-installer: host grammar needs improvements Bugster: 6786390 Issue: 2997 Enhancem.: gui-installer: disabled xmloutput file generation Enhancem.: gui-installer: removed remove host menu action in the installation Bugfix.: gui-installer: Fixed CSP mode prerequisites task failure Review: ZG LP-2009-04-16-1: Bugfix: error message when testing JAVA version Bugster: 6830972 Bugfix: load_sge_config.sh fails when userset pass and users not in a first interation Bugster: 6830081 Issue: 2994 Bugfix: Qmaster autoinstallation fails Review: MD LP-2009-04-16-0: Bugfix: Message from tar command appears in installation output Bugster: 6830900 Review: AP HP-2009-04-15-0: Bugfix: * Return values of SGE_Starter.exe are handled correctly now. * Logging of SGE_Starter.exe and SGE_Helper_Service.exe works now. Bugster: 6817190, 6817193 Issue: 2995, 2996 Review EB ZG-2009-04-14-0: Enhancem.: gui-installer: Improve table menu Bugster: 6786388 Bugfix: gui-installer: The result panel shows the admin/submit hosts as successfully installed even the last task failed. Bugfix: gui-installer - Continue feature: After successful BDB server installation the BDB host is added again by default to the host list. Bugster: 6830071 Bugfix: gui-installer: After resolving the host is not marked as RESOLVABLE but UNKNOWN_HOST even the IP is available just the execution of the arch script failed for some reason. Bugfix: 6830063 Review: LP LP-2009-04-14-2: Bugfix: gui-installer: first/last task might not be done on qmaster host Bugster: 6829933 Issue: 2991 Bugfix: gui-installer: install list might contain line with shadow and other component at once Bugster: 6829940 Issue: 2992 Bugfix: gui-installer: CopyExecutableCommand might fail due to an existing file Bugster: 6829947 Issue: 2993 Enhancem.: gui-installer: It should be possible to add admin submit hosts only Cleanup: gui-installer: commands handle spaces, added GetJvmLibCommand, cleanup Review: ZG LP-2009-04-14-1: Enhancem.: Improved comment in sgemaster script Review: ZG LP-2009-04-14-0: Bugfix: auto_install does not set the additional_jvm_args which limits qmaster size Bugster: 6827662 Issue: 3007 Bugfix: Existing local host configuration should be merged with the new one instead of replacing the old one Bugster: 6829839 Enhancem.: Adminuser JMX keystore generated during qmaster installation Enhancem.: Added keystore password length check for the auto install template Cleanup: JMX related install scripts Review: EB, AA AA-2009-04-14-0: Enhancem.: install logging.properties for sgeinspect Review: CR RC-2009-04-13-2: Enhancem.: refresh SPARC/Linux port (from Rayson) Review: RC RC-2009-04-13-1: Enhancem.: refresh NetBSD port (from Kevin M. Buckley) Review: RC RC-2009-04-13-0: Enhancem.: - Gentoo start-up script - MIPS/Linux port (both from Rayson) Review: RC LP-2009-04-09-1: Bugfix: Execd uninstallation should not remove host configuration if there is still a shadowd Bugster: 6828272 Issue: 2988 Review: EB LP-2009-04-09-0: Enhancem.: Improved timeout handling code reliability Enhancem.: improve hostname resolving error detection for invalid hostname settings Bugster: 6819732 Issue: 2983 Bugfix: sgemaster script might return 0 on failure Bugster: 6827877 Issue: 2987 Review: EB, AS ZG-2009-04-09-0: Enhancem.: gui-installer: New window to configure host specific settings as local execd spool dir, jvm path and args, connect user and timeouts. Review: LP LP-2009-04-08-0: Bugfix: IzPack CompilerConfig might report successful installer generation on unknown failure Review: ZG LP-2009-04-07-0: Enhancem.: gui-installer: Improved build process Review: ZG JG-2009-04-07-0: Bugfix: qmaster reports commlib read error in message file when qconf -m(s)conf fails Bugster: 6815435 Issue: 2953 Review: CR EB-2009-04-07-0: Doc: Added spec doc: Timed Event Thread Pool Review: JG, RD JG-2009-04-06-0: Bugfix: need tight integration accounting and sharetree usage according to reserved resources Bugster: 6746750 Issue: 2977 Review: RD RD-2009-04-06-1: Enhancem.: - corrected aimk -no-opt switch to really don't optimize - enabled optimization for aimk -debug switch Review: AA LP-2009-04-06-0: Enhancem.: gui-installer: Displaying (Sun) Grid Engine based based on distribution (sun/courtesy) Enhancem.: gui-installer: Reading SGE version from libs/gdi/version.c file Review: ZG RD-2009-04-06-0: Doc: enhanced man page for exclusive host scheduling Review: AA RD-2009-04-03-2: CVS: Created V62u3alpha2_TAG ----------------------------------------------------- LP-2009-04-03-2: Enhancem.: Renamed sgemacc to sgeinspect in install scripts Enhancem.: Improved auto install messages for shadowd Enhancem.: Validating JVM library in qmaster/shadowd installations Bugfix: qmaster startup might take too long if network is unavailable Bugster: 6824159 Issue: 2970 Bugfix: Qmaster does not display any message when started on non-qmaster host Bugster: 6824165 Issue: 2971 Review: AA, CR AA-2009-04-03-0: Bugfix: - Incorrect slots values for Queue Instance detail view - sgemacc->sgeinspect script and sgeinspect.zip - serialVersionUID for jgdi monitoring - sge_ca check for cacert.pem Bugster: 6824160 LP-2009-04-03-1: Bugfix: Fixing buildbreak on windows Review: RD RD-2009-04-03-1: Cleanup: some minor cleanups Review: CR LP-2009-04-03-0: Enhancem.: gui-installer: Enable RC should be displayed in shadowd installation Enhancem.: gui-installer: the start_gui_installer wrapper should check JAVA version Bugster: 6828239 Issue: 3012 Enhancem.: gui-installer: updated shadowd installation for changes in 6824933 Enhancem.: gui-installer: Improved checking for JVM library on qmaster and shadows when JMX enabled Bugfix: gui-installer: Execd spool dir check should check permissions also as root in root installation Bugfix: gui-installer: JVM library detection should offer a correct library based on qmaster architecture Bugster: 6786364 Issue: 2974 Review: ZG RD-2009-04-03-0: Bugfix: profiling output is wrongly accumulated Bugster: 6823086 Issue: 2975 Review: CR AP-2009-04-03-0: Bugfix: qrsub allows to request a non existing checkpoint object Bugster: 6755146 Issue: 2967 Review: JG LP-2009-04-02-1: Enhancem.: Added utilbin/valid_jvmlib to validate JVM library path before it's accepted in the installation Review: EB LP-2009-04-02-0: Enhancem.: Can install&enable JMX for running qmaster inst_sge -add-jmx Enhancem.: Upgrade in clone mode has a default PORT+10 instead of PORT+2 Enhancem.: Adminuser's keystore generated by default during qmaster installation (when JMX enabled) Enhancem.: Improved JVM library auto detection Enhancem.: JMX support for shadowd (initial version) Bugster: 6772999 Bugfix: CheckRunningDeamon for qmaster returns true even if no pid file created Bugster: 6824148 Issue: 2969 Bugfix: shadowd auto installation is not equivalent to the interactive installation Bugster: 6824933 Issue: 2973 Review: AP, AA, EB (6824933) ZG-2009-04-01-1: Enhancem.: Silent installation Bugster: 6786381 Enhancem.: Start-over feature on ResultPanel Enhancem.: Store the component selection with the name of the host in the file Enhancem.: Improve ResultPanel's content Bugster: 6786397 Enhancem.: JUnit test for class Host Bugfix: create test/classes diretory before compile Bugfix: On main progress bar, show the overall task count during installation Bugfix: Grammar fails on 'ge[1-]4' pattern. Bugfix: Timeout state is never shown at resolving Bugfix: Can enter loopback network interface name and address. Review: LP LP-2009-04-01-0: Reverting LP-2009-03-31-0 for TS LP-2009-03-31-0: Enhancem.: Can install&enable JMX for running qmaster inst_sge -add-jmx Enhancem.: Upgrade in clone mode has a default PORT+10 instead of PORT+2 Enhancem.: Adminuser's keystore generated by default during qmaster installation (when JMX enabled) Enhancem.: Improved JVM library auto detection Enhancem.: JMX support for shadowd (initial version) Bugster: 6772999 Bugfix: CheckRunningDeamon for qmaster returns true even if no pid file created Bugster: 6824148 Issue: 2969 Review: AP, AA MD-2009-03-31-1: Enhancem.: Added announce page, source snapshots LP-2009-03-30-0: Enhancem.: DetectJvmLibrary.jar available in util Review: AA CR-2009-03-30-0: Bugfix: qmaster csp installation fails when using port from /etc/services Bugster: 6784919 Issue: 2833 Review: MD RD-2009-03-26-0: Enhancem.: improved test_spooling_performance.sh Review: AS CR-2009-03-25-1: Cleanup: jsv_do_communication() did not return an answer list for every error Review: EB EB-2009-03-25-1: Cleanup.: made local variable in sge_peopen* non-static and added code comments Review: JG EB-2009-03-25-0: Enhancemn.: Added example section to jsv_script_functions man page Review: CR CR-2009-03-25-0: Bugfix: o qconf -m* reports misleading error if tmp dir is not writeable for the user and need 100% cpu Cleanup: o commlib: Fixed some errors reported by lint Bugster: 6819649 Issue: 2960 Review: RD RD-2009-03-24-0: Bugfix: - qconf -Mc did not print any output in case of success - build-in complex "display_win_gui" could be deleted Bugster: 6821207, 6821237 Issue: 2963, 2964 Review: CR EB-2009-03-23-0: Enhancemn.: JSV log command now has possibility to log empty lines Review: MD MD-2009-03-24-0: Enhancem: updated 62patches.txt file JG-2009-03-23-0: Bugfix: misleading execd logging "Can't get load values" on AIX make load_sensor protocol compliant Bugster: 6746960, 6842629 Issue: 2961, 3033 Review: DG DG-2009-03-19-0: Bugfix: stopping an execution daemon does not stop jobs Bugster: 6761993 Issue: 2759 Review: JG EB-2009-03-19-0: Bugfix: Some variable names can't be used in Bourne shell JSV scripts Bugster: 6818825 Issue: 2958 Bugfix: A qmaster not running a root exits with return code 1 if a JSV script to start as user is configured Bugster: 6810490 Issue: 2957 Review: JF SH-2009-03-19-0: Bugfix: qsub exits with a error msg when the event client was killed Bugster: 6808617 Issue: 2934 Reviewed by: DG ZG-2009-03-18-0: Enhancem.: Create JUnit test for HostPanel Enhancem.: Enable JMX by default Enhancem.: Option to build demo version Review: LP RD-2009-03-17-0: CVS: Created V62u3alpha1_TAG ----------------------------------------------------- RD-2009-03-17-0: Enhancem.: - added EXCL relop for boolean consumable Bugster: 6731441 Issue: 2629 Bugfix: - libspoolb and libspoolc should not link against jemalloc/mtmalloc Bugfix: - fixed memory leak in shepherd and buildin_starter Review: AA LP-2009-03-16-0: Enhancem: gui-installer build process improved Enhancem: updated the gui-installer version Review: ZG EB-2009-03-16-0: Bugfix: c_occasion job parameter can't be changed Bugster: 6814059 Issue: 2946 Bugfix: JSV: change of requested resources is rejected if also queues were selected during job submission Bugster: 6817691 Issue: 2954 Bugfix: influence of c_occasion on c_interval and vice versa is not documented explicitly Bugster: 6814060 Issue: 2947 Enhancemen.: minor change of sge_ca which allowes csp installaion in environments where root is not even mapped to nobody on NFS mounted fs.:wq Review: MD AH-2009-03-12-1: Bugfix: Incorrect referrences to "SGE_H_ALIASES" man pages Changed: hostnameutils(1), qping(1) Bugster: 6816705 Issue: 2056 Review: RD AH-2009-03-12-0: Bugfix: SIGUSR1 clash between s_rt and NOTIFY_SUS Changed: queue_conf(5) Bugster: 6816634 Issue: 2103 Review: RD MD-2009-03-11-0: Enhancem: problems with gaussian due to memory limitations of SGE Bugster: 6816037 Review: AS AA-2009-03-11-1: Bugfix: MACC_SRC_DIR absolute path fix Review: MD HP-2009-03-11-2: Bugfix: Fixed compile error HP-2009-03-11-1: Bugfix: Fixed wrongly merged inst_execd.sh file HP-2009-03-11-0: Bugfix: jobs started with -l dwg=true can't display their GUI on Windows Vista and Server 2008 Bugster: 6801748 Issue: 2904 Review: EB AA-2009-03-11-0: Bugfix: sgemacc distinst, comment in continuation lines Review: MD LP-2009-03-10-1: Enhancem: JMX used by default, -no-jmx option Review: AA LP-2009-03-10-0: Enhancem: Support for sgemacc Review: AA AA-2009-03-10-0: Enhancem: - qhost slots_resv - jgdi_qhost slots_resv - JobStateFilter corrections - jgdi.jar + jgdi-test.jar Bugster: 6840435 Issue: 3028 Review: RD, RH RD-2009-03-10-0: Bugfix: fixed qmaster abort at update from previous version Bugster: 6814823 Issue: 2949 Review: JG GH-2009-03-10-0: Bugfix: qmod -r does not delete the old instance of a job before the new one is started Bugster: 6244911 Issue: 1521 Review: JG JF-2009-03-10-0: Bugfix: Fix qsub truncation of scripts Issue: 2876 Bugster: 6810688 Review: AP RC-2009-03-07-1: Enhancem: I/O accounting on Linux Issue: 2429 Review: RC (original patch by johna) RC-2009-03-07-0: Cleanup: - fixed build on lx-ppc - removed dependency on ssl & crypto libs when SGE is built with -no-secure MD-2009-03-06-0: Enhancem: updating webpages for 62u2 courtesy binaries EB-2009-03-06-0: Bugfix: JSV -w param has wrong value inside script Bugster: 6806219 Issue: 2917 Bugfix: Changing "w" job parameter to "n" in client JSV is not handled correctly Bugster: 6810493 Issues: 2938 Bugfix: DRMAA test_drmaa_perf produces segmentation fault when using user .sge_request file Bugster: 6806363 Issues: 2923 Bugfix: qsub man page refers to "see -jsv option below" but it is above in some cases Bugster: 6806324 Issue: 2939 Bugfix: The cluster config paramter jsv_allowed_mod does not work -w switch Bugster: 6806209 Issue: 2940 Review: CR ZG-2009-03-04-0: Enhancem.: Major cleanup: -new package hierarchy as 3party/IzPack/src/lib/com/... -new package .../grid/installer/task -move nested classes to separate files -redesign progress bar refreshing -generalize worker threads -write comments Bugster: 6812970 Review: LP DG-2009-03-04-0: Cleanup: Added Java DRMAA issue test. Bugster: 6796175 Issue: 2882 Review: CR ZG-2009-03-03-0: Enhancem.: Update man pages for v62u2_tag CR-2009-03-03-0: Enhancem.: test_general binary now prints FD_SETSIZE value Review: DG RD-2009-02-26-0: Bugfix: fixed scheduling issue with fill_up allocation rule Bugster: 6807072 Issue: 2901 Review: CR MP-2009-02-25-0: Bugfix: Fixed temp file name in loadsensor has problems on arch AIX. Merged from MP-2009-02-23-0 Bugster: 6740546 Issue: 2715 Review: DG RD-2009-02-19-0: CVS: Created V62u2_TAG ----------------------------------------------- LP-2009-02-19-0: Bugfix: gui-installer: cancel button does not work during installation gui-installer: timeout values not correctly enforced when machine is overloaded Review: ZG LP-2009-02-17-1: Bugfix: - fixed no newline at end of file in GUI installer help pages Review: ZG LP-2009-02-17-0: Bugfix: - fixed no newline at end of file in GUI installer help pages Review: TBD RD-2009-02-17-0: Bugfix: - disabled jemalloc for lx24-x86 and lx24-ia64 - fixed no newline at end of file in GUI installer templates Review: JG ZG-2009-02-16-0: Bugfix: Update docs DG-2009-02-16-0: Bugfix: Installation of execd on AIX will fail on filesystems with root to nobody mapping when trying to copy loadsensor script automatically. Review: HP JG-2009-02-16-0: Bugfix: qrsh (new interactive job support) does not work with host aliasing Bugster: 6805914 Issue: 2915 Review: RD LP-2009-02-16-0: Bugfix: gui-installer: Installation might fail for C shell users Bugfix: gui-installer: Windows installation via rsh does not work Bugfix: gui-installer: Infinite loop when installing win execd host due to missing LOCAL_EXECD_SPOOL Belongs to: LP-2009-02-13-1 Bugster: 6808084 Issue: 2927 Bugfix: gui-installer: Install button not enabled when going back and forth Bugfix: gui-installer: ant clean target does not delete dist and build directories Bugfix: gui-installer: When installing qmaster with shadow daemons, it should be shadowd too Bugfix: gui-installer: Windows hosts should have a suggested default local spool dir Bugfix: gui-installer: Install progress bar goes backwards for a while Bugfix: gui-installer: Log might be empty due to '!' in command output Enhancem: gui-installer: Increased default timeout values Enhancem: gui-installer: Support for windows connect_mode Review: ZG CR-2009-02-16-0: Bugfix: o sge_execd may crash at start-up time due to fd_set used incorrectly in sge_close_all_fds() Bugster: 6288945 Issue: 505 Bugfix: o jsv_is_send_ready() now uses poll() system call when USE_POLL is defined Cleanup: o removed not used functions in sge_execd Enhancem.: o new utilib function: sge_get_max_fd() Review: RD EB-2009-02-16-0: Doc: added new manpage describing the JSV script interface Review: SK ZG-2009-02-15-2: Bugfix: gui-installer: rcp does not support '-l' switch to specify user Review: LP LP-2009-02-13-2: Bugfix: gui-installer: Updated help pages to reflect latest changes Review: ZG ZG-2009-02-13-0: Bugfix: gui-installer: have to be able to change the user name to adopt to Windows hosts naming convention Bugfix: gui-installer: have to distinguish between normal and Windows host installation to apply domain name prefix to user name Bugfix: for Windows host the local execd spool dir is necessary Bugfix: check admin user's permissions regarding to SGE_ROOT after Main Configuration panel Review: LP LP-2009-02-13-1: Bugfix: gui-installer: Infinite loop when installing win execd host due to missing LOCAL_EXECD_SPOOL Bugfix: gui-installer: Installing windows hosts fail due to wrong connect user (introduced windows mode) Review: ZG DG-2009-02-13-0: Bugfix: Fixed Java DRMMA bug when parsing native specification with empty string. Belongs to: RD-2009-02-05-0 Bugster: 6802994 Issue: 2910 Review: RD LP-2009-02-13-0: Bugfix: gui-installer: CSP installation will fail if SGE_CELL is set before starting gui-installer Bugfix: gui-installer: CSP installation possible only with rsh (ssh should work too) Bugfix: gui-installer: Install log might contain only a message /tmp/install.6554 is missing Enhancem.: gui-installer: Support for connect_user Bugfix: gui-installer: Works properly only with sh or bash as default root/connect_user shell Bugfix: gui-installer: UI fixes Review: ZG AA-2009-02-12-0: Bugfix: qmon: duration and end_time fields in Advance Reservation Window are both effective Spinbox problems on lx26-amd64 reverting mem leak fix which leads to crash Bugster: 6804626 Review: RD ZG-2009-02-12-1: Bugfix: Invalid parsing of installer.jar arguments Bugfix: Spooling method gets modified after checking hosts Bugfix: Qmaster is always in the shadow host list Bugfix: Time-out state is missing Enhacem.: Shadow component is unchecked by default Enhacem.: Admin component is unchecked by default Review: LP RD-2009-02-12-1: Bugfix: fixed jemalloc deadlock with old linux pthread implementations Review: JG HP-2009-02-12-0: Bugfix: fixed wrong description of parameters of -j option in qacct(1) man page Review: JG RD-2009-02-12-0: Bugfix: fixed more qmon memory leaks Belongs to: RD-2009-02-11-1 Review: AA RD-2009-02-11-4: Bugfix: fixed parsing warning in jsv(1) Review: JG RD-2009-02-11-3: CVS: Created V62u2_RC2_TAG --------------------------------------------------- EB-2009-02-11-0: Bugfix: Messages on JSV stderr stream are now logged to message file but next job is not rejected anymore if the message occured before the START command was sent. Review: MD ZG-2009-02-11-0: Bugfix: './start_gui_installer -help' does not terminates Bugfix: fix cancel button behaviour Bugfix: on NFS the root user does not have permission to write into cell directory RD-2009-02-11-1: Bugfix: fixed qmon memory leaks Review: AA RD-2009-02-11-0: Bugfix: jemalloc was not effective because it was too late in the linker command line Review: JG CR-2009-02-11-0: Bugfix: fixed sge_shadowd core dump when it has to start a new qmaster daemon Belongs to: AA-2009-01-14-0 Review: MD LP-2009-02-10-0: Bugfix: gui-installer: check_host might always fail due to invalid architecture Bugfix: gui-installer: installation will fail on different architectures from the current host Enhacem.: gui-installer: can change thread pool sizes and resolve/install timeout values as arguments Review: ZG RD-2009-02-10-0: Bugfix: fixed memory leaks and access to uninitialized variables Review: AA EB-2009-02-10-0: Cleanup: Removed debug code and added ADOC comments Review: AS, MD RD-2009-02-09-1: Bugfix: spooledit can not handle lines greater than 255 chars Bugster: 6802883 Issue: 2908 Review: JG RD-2009-02-09-0: Enhancem.: - reduced queue instance sorting during dispatch runs - improved hash recreation necessary after list sorting - fixed build dependency for test_performance Review: JG LP-2009-02-06-0: Bugfix: inst_sge -copycerts in AUTO mode does not work Bugster: 6808089 Issue: 2928 Bugfix: gui-installer: IPs treated as hostnames Bugfix: gui-installer: prerequisites/final tasks now displayed in the install list Bugfix: gui-installer: Can't add only admin/submit hosts Bugfix: gui-installer: admin/submit only hosts should not skipped by the installer Bugfix: gui-installer: Many errors not discovered when rsh is used (rsh always returns 0) Bugfix: gui-installer: Removed unnecessary fields in CSP certificate screen Bugfix: gui-installer: admin/submit host list in the final readme may be incomplete Bugfix: gui-installer: CSP panel shown in incorrect scenarios Bugfix: gui-installer: ProgressBar sometimes not shown Enhancem: gui-installer: Improved task timeout times Enhancem.: gui-installer: Improved installation log formatting Enhancem.: gui-installer: check_host does not need to be done for admin/submit only hosts Enhancem.: gui-installer: Improved installation panel UE Cleanup: gui-installer: Util and CommandExecutor Review: ZG JG-2009-02-06-0: Bugfix: qrstat -help shows advance reservations Bugster: 6801151 Issue: 2903 Review: RD CR-2009-02-05-0: Bugfix: module test test_issues (commlib) might fail on fast hosts Review: AH RD-2009-02-05-1: Bugfix: fixed critical scheduler message logged wrongly Review: DG RD-2009-02-05-0: Bugfix: - fixed buffer overrun in drmaa - test_drmaa did not executed drmaa_exit in case of success Belongs to: JG-2009-01-28-0 Bugster: 6802994 Review: DG RD-2009-02-03-1: CVS: Created V62u2_RC1_TAG --------------------------------------------------- LP-2009-02-03-1: Bugfix: gui-installer: SMF not used at all Bugfix: Autoinstallion on loopback interface ends with exit 0 Bugster: 6800497 Issue: 2929 Review: AA LP-2009-02-03-0: Bugfix: gui-installer: admin/submit hosts not set properly Bugster: 6800509 Bugfix: gui-installer: CSP installation does not work properly Bugster: 6790504 Review: ZG ZG-2009-02-03-0: Bugfix: gui-installer: it's not possible to switch-off 'service automatic startup' feature Bugster: 6793169 Bugfix: in case of standalone execd host installation the install table model fails due to a fireTableRowsUpdated event on a row with index -1. Bugfix: 'losing' reference to a host which already exists in the list Bugfix: clear table selections before checking/installing hosts Bugfix: disable tables during host checking to avoid inconsistent modifications Bugfix: in check_host script if there is no local execd spool dir specified check permissions of the global execd spool dir Enhancem.: let the user to set the thread pool sizes in start_gui_installer script Review: LP EB-2009-02-03-0: Bugfix: TCL server JSV script does not work on some architectures Bugster: 6785207 Issue: 2836 Review: TBD RD-2009-02-03-0: Enhancem.: - removed gdi dependency from test_performance - added script to measure spooling performance easily Review: JG RD-2009-02-01-0: Enhancem.: optimized scheduler event subscriptions for lower memory consumption Review: JG CR-2009-02-02-1: Bugfix: hostname resolving problem might cause execution daemon shutdown Issue: 2895 Bugster: 6799028 Review: RD CR-2009-02-02-0: Bugfix: using IP String in first column of host_aliases file produces dead lock Issue: 2897 Bugster: 6799088 Review: RD ZG-2009-01-30-0: Bugfix: gui-installer: Missing Bdb spool directory field in case of root user. Bugster: 6799969 Bugfix: gui-installer: during installation the qmaster's installation task is not shown on the progress bar Bugfix: gui-installer: on ResultPanel the "Save as..." button pops-up an open file dialog Bugster: 6799964 Bugfix: gui-installer: after resolving the problem discovered by the check_host script the state does not get modified to "Reachable" Bugfix: gui-installer: the scroll bar on log window is positoned to the bottom of the text area Bugfix: gui-installer: check_host script does not return back with the proper exit value in case of rsh. Bugster: 6799971 Enhancem.: change "Resolve" button to "Add" for better understanding Enhancem.: "No" button is the default in case the check_host fails on some host and the user is asked whether continue the installation Enhancem.: new host state for installation failure due to dependency on previously failed component Enhancem.: new host state for installation failure due to already installed component Enhancem.: new introduced confirmation dialog to change qmaster selection on Host Selection panel Enhancem.: improved tooltips LP-2009-01-30-1: Enhacem.: gui-installer: start_gui_installer.sh renamed to start_gui_installer in distribution Enhacem.: gui-installer: start_gui_installer showns no messages unless -debug is used Review: ZG LP-2009-01-30-0: Bugfix: gui-installer: grammar improvements + tests o IPs added as hostnames o hostnames with "_" not recognized o "-" in a hostname could be treated as a range Bugster: 6786390 Bugfix: gui-installer: installation can be started on host with no component selected Bugfix: gui-installer: host lists could contain same host multiple times Bugfix: gui-installer: EXEC_HOST_LIST_RM not filled Bugfix: gui-installer: qmaster install timeout increased to 2 mins Bugfix: gui-installer: execd-spool-dir check must be done just for admin user Review: ZG RH-2009-01-30-0: Bugfix: JGDIProxy sets it's own classloader into the JMX enviroment Review: AA RD-2009-01-29-0: Bugfix: fixed use of default duration for unbound jobs Bugster: 6799032 Issue: 2896 Bugfix: fixed commlib write thread timeout for ssl framework Belongs To: RD-2009-01-28-0 Enhancem.: some smaller scheduler improvements Bugfix: fixed buffer overflow in spooledit Review: AA EB-2009-01-29-0: Bugfix: read_raw doesn't return 1 on EOF Bugster: 6788514 Issue: 2849 Bugfix: qalter -h u fails if a jsv_url is configured Bugster: 6788539 Issue: 2850 Bugfix: insufficient error message from JSV if script fails Bugster: 6788570 Issue: 2852 Bugfix: master unresponsive if jsv script gets killed (only in MAINTRUNC) Review: MD CR-2009-01-28-0: Bugfix: fixed compilation for aimk -no-secure Review: AA JG-2009-01-28-0: Bugfix: drmaa: quotes in native specification can result in memory corruption Bugster: 6317028 Issue: 1768 Bugfix: Some minor memory leaks and cleanup. Review: DG RD-2009-01-28-0: Bugfix: fixed commlib write thread timeout Related To: CR-2008-12-08-0 Review: CR ZG-2009-01-27-0: Bugfix: in case a host is added by pressing 'Enter' host lists lose their consistency Enhancem.: expand the size of the window to 940x720 Enhancem.: improve tooltips Review: LP RD-2009-01-26-2: Bugfix: fixed minor memory leak in cull error handling Review: JG RD-2009-01-26-1: CVS: Created V62u2beta2_TAG ---------------------------------------------------- CR-2009-01-26-0: Bugfix: Internal event clients, like the scheduler, do not reconnect after timing out Bugster: 6793076 Issue: 2890 Cleanup: o smaller commlib cleanups Enhancem.: o disabled call of application info callback function when SIM clients name is not "qping" Review: RD RD-2009-01-26-0: Bugfix: fixed huge qmaster memory usage after hostgroup modification Bugster: 6797579 Issue: 2887 load_sge_config.sh does not work huge amount of configuration objects Bugster: 6797581 Issue: 2888 introduced scheduling performance regression with Belongs to: RD-2008-11-03-0 ZG-2009-01-26-0: Bugfix: gui-installer: Should warn a user that some of the selected components were not really selected Bugster: 6786371 Bugfix: gui-installer: it's not possible to switch-off 'service automatic startup' feature Bugster: 6793169 Bugfix: gui-installer: should check bdb server spool dir permissions on the remote host Bugster: 6786338 Bugfix: gui-installer: qmaster spool dir permission check done locally Bugster: 6786355 Bugfix: gui-installer: GUI installer should generete auto_inst file before starting with the installation Bugster: 6786384 Enhacem.: Improve debugging capabilities Enhacem.: Check the FS type of BDB spool directory remotely Review: LP GH-2009-01-26-0: Bugfix: Subordinate suspension state can not be changed Bugster: 6793687 Issue: 2886 Review: DG LP-2009-01-22-0: Bugfix: gui-installer: Parallel shadowd installation might fail Bugfix: gui-installer: Detecting missing arch script and timeout during host resolution Bugfix: gui-installer: Requires passwordless connections from each execd to the qmaster host Bugfix: gui-installer: check_host script not copied to the distribution Enhacem.: Improved confusing error message about cell_name Bugster: 6786392 Review: ZG GH-2009-01-22-0: Bugfix: bad logging if "act_qmaster" file does not exist Bugster: 6288964 Issue: 2881, 860 Review: CR AH-2009-01-22-0: Bugfix: Sequence numbers broken for parallel jobs when multpile queues are used Changed: libs/sched/sge_select_queue.c Issue: 2864 Bugster: 6791556 Review: RD DG-2009-01-22-0: Bugfix: Calling drmaa_exit before any calls to drmaa_init results in segfault Issue: 2882 Bugster: 6796175 Review: AA RD-2009-01-21-0: Bugfix: - subordinate list generated a critical message entry - subordination was broken when setup was done in wrong order Bugster: 6795248, 6796068 Issue: 2883, 2884 Review: CR DG-2009-01-21-0: Cleanup: aimk -only-depend does not work on linux Review: AA JG-2009-01-19-0: Bugfix: end of parallel job doesn't kill subsequently added, longer running tasks Bugster: 6579326 Issue: 2329 Review: RD DG-2009-01-19-0: Bugfix: execd: "ptf complains: Job does not exist" message in log file Merged from: V61_BRANCH (DG-2009-01-15-0) Bugster: 6326191 Issue: 1806 Review: JG JG-2009-01-15-0: Bugfix: no scheduler activity to schedule pending jobs when the qmaster installed on RHEL4 Bugster: 6792704 Issue: 2877 Bugfix: qmaster param SCHEDULER_TIMEOUT is used for all event clients Bugster: 6794303 Issue: 2878 Review: CR AA-2009-01-14-0: Bugfix: - qmon misses Poke (-w p) in "Verify Mode" - adding hostname="u*" results in error: can't resolve host u*. Works on command line - max_advance_reservations is missing - wrong certificate path for jmx with /etc/services Issue: 2874,2875 Bugfix: 6793547,6793554 Review: RD ZG-2009-01-13-0: Enhancem.: Execute directory permission checkings on remote host Enhancem.: Switch-on logging by default Bugfix: Disable "Browse" button in case the qmaster is not on localhost Bugfix: New port validator for JMX port value Review: LP AP-2009-01-13-0: Bugfix: project(5) man page should say that "acl" set to "NONE" means "all users" Bugster: 6784461 Issue: 2827 Bugfix: man sched_conf minor corrections Bugster: 6784258 Issue: 2873 Review: MP RD-2009-01-12-0: Bugfix: qsub -w e crashes qmaster if schedd_job_info is enabled Bugster: 6749902 Issue: 2742 Review: CR AP-2009-01-12-0: Bugfix: broken output with qstat -r when a job requests more than a single queue using -q wc_queue_list Bugster: 6732134 Issue: 2682 Review: JG Bugfix: shell_start_mode should be documentated to be only used for batch jobs Bugster: 6792604 Issue: 2058 Review: DG JG-2009-01-12-0: Bugfix: qing aborts when connected to an execd running a tightly integrated parallel job Bugster: 6791931 Issue: 2868 Review: RD DG-2009-01-12-0: Bugfix: man queue_conf section on subordinate list is a little unclear Bugster: 6344900 Issue: 2732 Review: JG EB-2009-01-09-0: Bugfix: qalter rejects other modification if -w is not explicitly allowed in jsv_allowed_mod when JSV is used Issue: 2854 Bugster: 6788644 Bugfix: qalter rejects modification of hold state if -t is not also allowed in jsv_allowed_mod when JSV is used Issue: 2855 Bugster: 6788646 Bugfix: qresub does not set a new job id for the new job. Issue: 2839 Bugster: 6786343 Bugfix: Tight pe job might stuck in system although enforce_limit parameter is set. Issue: 2856 Bugster: 6788647 Bugfix: echo_raw -r without additional arguments produces core dump Issue: 2848 Bugster: 6788512 Bugfix: Impossible to change -i -e and -S via JSV Issue: 2857 Bugster: 6788649 Bugfix: Impossible to delete -ckpt via JSV if -c was still set Issue: 2858 Bugster: 6788665 Bugfix: Pe range (pe_min, pe_max) can't be modified even if it is specified in jsv_allowed_mod list when JSV is used Issue: 2859 Bugster: 6788651 Bugfix: Master rejects job if JSV verification process takes longer than 10 seconds. Issues: 2860 Bugster: 6788652 Bugfix: Documentation of JSV scripts is missing. Issue: 2861 Bugster: 6788654 Review: MD ZG-2009-01-08-0: Update: update html man pages Review: LP MP-2009-01-05-0: Bugfix: qstat -j XXX -xml schema not well documented. Bugster: 6320659 Issue: 1776 Review: JG JG-2009-01-02-0: Bugfix: incomplete accounting for the last short tasks of tightly integrated parallel job Bugster: 6728379 Issue: 2815 Bugfix: aborted slave task on tight integration job generates error log in execd Bugster: 5082660 Issue: 1197 Bugfix: qmaster reports commlib read error in message file when qstat -j is executed Bugster: 6786406 Issue: 2840 Review: DG DG-2008-12-30-0: Bugfix: JSV(1) man page issues Issue: 2863 Bugster: 6789620 Review: JG JG-2008-12-29-0: Bugfix: sge_pe man page should document that job_is_first_task is evaluated if wallclock accounting is configured Issue: 2862 Bugster: 6741415 Review: DG AP-2008-12-23-0: Bugfix: o shepherd_cmd is missing from sge_conf(5) man page o man qsub has wrong formatting in -C section o qconf -Msconf filename at wrong position in man page o man qconf should mention sge_resource_quota(5) in "SEE_ALSO" Bugster: 6575743, 6769528, 6788577, 6788579 Issue: 2318, 2770, 2215, 2389 Review: JG JG-2008-12-23-0: Bugfix: the man page accounting(5) is missing info about 2 new fields introduced in 6.2 Issue: 2853 Bugster: 6778386 Review: AP JG-2008-12-22-0: Bugfix: sge_resource_quota(5) should warn that resource quotas are not enforced as job resource limits Issue: 2793 Bugster: 6773584 Review: MD RD-2008-12-19-0: Bugfix: TMPDIR of a tight integration job is sometimes not deleted Issue: 2846 Bugster: 6787485 Review: CR CR-2008-12-19-0: Bugfix: o fixed test_cl_commlib module test binary which might left deadlocked processes if used with CL_FORK_SLEEP_CHILD environment variable. Belongs to: CR-2008-12-18-0 Review: RD JF-2008-12-18-0: Enhancem.: Improvement of existing man pages Bugster: 6609691 Issue: 0 Bugster: 6575730 Issue: 0 Review: MP RD-2008-12-18-0: Bugfix: - improved reprioritize data structures reprioritize code is no longer executed if feature is disabled Cleanup - several code improvements Bugster: 6786957 Issue: 2843 Review: CR CR-2008-12-18-0: Bugfix: o execd job start might produce communication errors because of not posix compliant fork() system call Cleanup: o poll()/select() error handling Bugster: 6786152 Issue: 2837 Review: RD ZG-2008-12-18-0: Enhancem.: new intermediate page for manual version selection MP-2008-12-17-0: Bugfix: submit(1) man page defines support for the -display option only for qsh, however qrsh also supports this Bugster: 6785777 Issue: 2841 Review: AP ZG-2008-12-17-1: Update: update html man pages ZG-2008-12-17-0: Bugfix: mk_ascman: does not generate the '?pathrev=' string to the end of the links inside man pages Revert: mk_ascman: don't show the sge versions on the left side frame at the top Review: LP LP-2008-12-15-3: Bugfix: gui-installer - permission validator fixed Bugfix: gui-installer - fixed ignore domains field Review: ZG LP-2008-12-15-2: Bugfix: gui-installer - Missing content in html help pages Review: ZG ZG-2208-12-15-0: Enhancem.: Increase the number if concurrent install threads to 8 Bugfix: Wrong permission checking Bugfix: Let the user set the HOSTNAME_RESOLVING value Bugfix: In case of standalone BDB server installation don't check the directory of the cell Enhancem.: Related to prev. bugfix; Improve rule engine to handle brackets. Review: LP JF-2008-12-15-0: Enhancem.: Improvement of existing man pages Bugster: 6518002 Issue: 0 Bugster: 6784558 Issue: 0 Review: MP LP-2008-12-15-1: Cleanup: Removed debug output from the install_component LP-2008-12-15-0: Several gui-installer bugfixes Bugfix: Standalone execd installation does not work if not started from qmaster host Bugfix: Remove hosts does not remove host resolved hosts Bugfix: Local commands should be executed locally (without a remote shell) Bugfix: Install button may remain disabled when all hosts are removed Bugfix: It's possible to start installation when no component are selected Bugfix: Readme generated without a timestamp Bugfix: Readme generated only for a default cell Enhacem.: Shadow daemons not installed by default Review: ZG RD-2008-12-12-1: CVS: Created V62u2beta_TAG --------------------------------------------------- ZG-2008-12-12-0: Bugfix: turn off bold fonts in case of Metal look-and-feel Bugfix: hide error label on host selection panel when input type gets modified Bugfix: show execd spool dir in host selection table only if the host signed as execd host Bugfix: after deletion of qmaster/bdb host it's not possible to sign other hosts as qmaster/bdb Review: LP RD-2008-12-12-0: Bugfix: fixed execd crash with tight integrated jobs Review: CR Bugster: 6784528 Issue: 2831 RD-2008-12-11-1: CVS: Created V62u2alpha3_TAG ----------------------------------------------------- LP-2008-12-11-0: Bugfix: ADMIN_USER not used in auto install mode Bugster: 6808600 Issue: 2930 Bugfix: When all.q or allhosts already exists qmaster installation fails Bugster: 6673742 Issue: 2594 Bugfix: ShutdownQmaster - can't use SMF when not root Bugster: 6808604 Issue: 2931 Bugfix: Removed No CSP System installed message in auto installation Bugfix: Fixed gui-installer, test, javadoc, clean targets Bugfix: gui-installer - remove existing qmaster does not uninstall execds first and fails Bugfix: gui-installer - final readme page not generated Bugfix: gui-installer - SHELL_NAME rsh does not work Bugfix: gui-installer - Using /tmp as temp dir on all ARCH Bugfix: gui-installer - fstype checks non-existing dir Bugfix: gui-installer - Should first install BDB, qmaster, and only then the rest Bugfix: gui-installer - Standalone bdb installation calls qconf and fails bacause of that Review ZG, AA CR-2008-12-11-1: Bugfix: - fixed possible core dump for qsub -jsv in sge_gdi_packet_execute_external() when job verify fails Review: RD RD-2008-12-11-0: Bugfix: - platform was not correctly regonized on Vista - corrected CSP directory name handling - added check for execed_spool_dir name at installation Bugster: 6783373, 6783850, 6783835 Issue: 2824, 2825, 2826 Review: CR JG-2008-12-11-0: Bugfix: tightly integrated jobs: replaced client cache by environment variable SGE_RSH_COMMAND set by execd Bugster: 6695481 Issue: 2393 Review: RD ZG-2008-12-11-0: Bugfix: new label for Berkeley db server text field Bugfix: hide the 'Remove existing component' checkbox and feature Bugfix: missing execd spool dir value at qmaster host added by default to host selection table Bugfix: correcion of error label behaviour on HostPanel Bugfix: in some cases the execd spool directory value does not gets substituted Bugfix: qmaster has value 'none' in case of standalone execd/shadowd installation Bugfix: get the FS type of the existing parent Enhancem.: custom icon and red text color for the error label on HostPanel Enhancem.: host resolving by pressing 'Enter' button Review: LP CR-2008-12-11-0: Cleanup: final cleanup of poll() enhancements o added additional error handling if a poll() error POLLHUP, POLLERR or POLLNVAL o re-worked error return when read() returns zero data length which might be possible when using poll() o reset is_xxxx_selected flags in connection struct also if poll()/select() returns 0(timeout) or -1(error) so that connection garbage collector can remove them also o fixed test_virtual_gdi_client Belongs to: CR-2008-12-08-0 Review: RD MD-2008-12-10-0: Bugfix: qconf -bonsai option not mentioned in -help or man page Bugster: 6373881 Review: CR ZG-2008-12-10-0: Bugfix: admin user field should show up only at qmaster installation Bugfix: qmaster host field should show up only at qmaster installation Bugfix: cell name (directory) should be validated regarding to sge_root. Bugfix: qmaster and execd port fields should show up only at qmaster installation Bugfix: range validator for qmaster and execd port values Bugfix: shell and copy fields should show up at qmaster or execd installation Bugfix: add.to.rc field should be invisible in case of non-root installation Bugfix: permission validator for jmx keystore directory Bugfix: permission validator for qmaster and execd spool directories Bugfix: validate country code filed Bugfix: check permissions on jmx keystore dir in case of express installation Bugfix: fix 'file' field orientation on UserInputPanel Bugfix: check checkbox and radiobutton dependencies on UserInputPanel Enhancem.: permission validator Enhancem.: method to source settings.sh in Util Enhancem.: method to source act_qmaster in Util Enhancem.: method to get user's group in Util Enhancem.: add capeabality for ExtendedFile to check user permissions Enhancem.: source settings.sh in case of express and execd or shadow host installation Enhancem.: add validator capability to 'file'field on UserInputPanel Review: LP EB-2008-12-10-0: Enhancemn.: - Added test functionality to qmaster to slow down worker threads. Used in TS for certain tests in combination with JSV Bugfix: - ac job parameters where exproted to JSV as environment variabes and not as parameters. - fixed bug is bourne shell JSV which made it impossible to add JSV parameters if they where not previously expored by the master. - fixed bug which rejected first submitted job after JSV process was kill or when a protocoll error occured. - changing the checkpointing interval via JSV was not handled correctly - read write lock was hold during submission if JSV was triggered - sleep ind JSV blocked worker thread kill and restart was not handled correctly - JSV was started with wrong permission if if no user or admin user was specified as start user MP-2008-12-09-0: Bugfix: max_unheard default value in sge_conf man page shows incorrect value Bugster: 6399453 Issue: 2268 Review: JG RD-2008-12-09-0: CVS: Created V62u2alpha2_TAG ----------------------------------------------------- LP-2008-12-09-1: Bugfix.: Installation broken. CopyCA always performed. Review JG LP-2008-12-09-0: Enhancem.: Reverted default to no-jmx (SGE 6.2 behavior) Bugfix: gui-installer did not use csp/afs flags Cleanup: gui-installer cleanup for no-jmx as default Review JO LP-2008-12-08-0: Bugfix: qmaster auto installation with jmx will fail if CSP_RECREATE flag is not set Bugster: 6808597 Issue: 2932 Following changes apply to clients/gui-installer Cleanup: Removed install key shortcut Bugfix: inst_sge - fixed variable settings for -autogui Bugfix: JVM lib still not correctly detected on MacOS Bugfix: CSP panel showned when JMX SSL selected Bugfix: Command executor output now reliable Bugfix: FSType commands not executed on remote host Bugfix: GUI installer files should be referenced against $SGE_ROOT, not the work dir. Enhancem.: Commands executed locally if on local host Enhancem.: COPY_COMMANDS get skipped when on local host Enhancem.: Added ssh/scp options to supress TTY outputs Enhancem.: All install tasks fail once it's first (bdb or qmaster) task gets terminated due to timeout Review ZG, EB CR-2008-12-08-1: Bugfix: fixed module test binary test_cl_commlib_client (wrong shutdown if CL_RUNS was set) Review: AA CR-2008-12-08-0: Bugfix: Commlib: Use poll() instead of select() for linux and solaris architectures Commlib: commlib logging output now also contains os thread id Belongs to: CR-2008-11-28-0 Review: RD AA-2008-12-08-0: Bugfix: several auto installation problems with jmx mode - installation of certificates in autoinstall even if -no-jmx is set - non-root user installation fails due to wrong keystore path - installation failure in auto install mode if JAVA_HOME is not set - inst_template.conf missing additional jvm args -Xmx=256m - auto install broken if SGE_JMX_PORT not set correctly even if in -no-jmx mode Bugster: 6782086 Issue: 2823 Review: EB AA-2008-12-05-0: Enhancem.: - JGDI methods getSgeQmasterPort(), getSgeExecdPort() - JGDIProxy method getMBeanServerConnection() Review: EB MP-2008-12-05-0: Bugfix: The sge_conf man page does not contain any information regarding SCHEDULER_TIMEOUT Bugster: 6288944 Issue: 498 Review: JG JG-2008-12-04-0: Bugfix: - qsub(1) explanation of -b n shall mention performance impact - load sensor information is incomplete in sge_execd(8) Bugster: 6535332, 6575742 Issue: 2208, 2291 Review: MP MD-2008-12-04-0: Bugfix: o qconf -kej should remove jobs and then shutdown execd o description of qsub -notify option is inadequate Bugster: 6675023, 6704958 Review: DG RN-2008-12-03-0: Bugfix: qmaster exits with lGetString: wrong type for field JG_qhostname (lHostT) Bugster: 6776137 Issue: 2816 Review: JG EB-2008-12-04-0: Enhancemn.: - additional JSV parameter in global conf which makes it possible to reject the use of certain qalter switches if server side JSV is used - corresponding changes in qmon man page - added missing jsv_url description to sge_conf. - fixed bug in read_raw implementation which might cause hanging client JSV scripts on some architectures. Review: MD JG-2008-12-03-0: Bugfix: fixed memory leaks Review: DG ZG-2008-12-03-0: Enhancem.: improve mk_ascman script the way to generate html pages which handle multiple sge versions Enhancem.: update html man pages Review: LP LP-2008-12-02-0: Enhancem.: autoconf doesn't need JMX args for execd uninstall Bugster: 6808594 Issue: 2933 Following changes apply to clients/gui-installer Bugfix: jvm library name now resolved correctly on MacOS Bugfix: SSL enabled by default for JMX Bugfix: Added default JMX flag Bugfix: Java target and source 1.5 enforced Enhancem.: Qmaster installation allowed for any host, but win Enhancem.: New aimk switches -no-gui-inst and -only-gui-inst Enhancem.: Improved JMX SSL labels Enhancem.: Shadowd, Execd installation does not start if bdb, or qmaster installation fails. Review ZG EB-2008-11-28-0: Bugfix: - fixed erros in jsv man page Review: MD RD-2008-11-28-1: CVS: Created V62u2alpha1_TAG ----------------------------------------------------- CR-2008-11-28-0: Cleanup: Commlib: Use poll() instead of select() for linux and solaris architectures Review: RD RD-2008-11-28-0: Bugfix: - added SISSL header to jsv.pl - corrected typos in aimk LP-2008-11-28-0: Cleanup: Added missing headers, support for build host Review ZG JG-2008-11-28-0: Bugfix: Fixed memory leaks in JSV and qmaster JVM thread. Review EB LP-2008-11-27-1: Cleanup: GUI installer, license handling Review MF TB-2008-11-27-0: Bugfix: renaming of commands in JSV Perl script to match TCL names Review: MD LP-2008-11-27-0: Enhancem.: Gridengine GUI Installer - initial version Review ZG, JO, RD, MD RD-2008-11-26-0: Enhancem.: enabled jemalloc as default for linux hosts Review JG EB-2008-11-25-0: Bugfix: - removed deadlock situation in qmaster in combination with server JSV - fixed bug in Perl JSV Review: RD LP-2008-11-24-0: Enhancem.: Modifed JMX behavior -jmx and no flag are equal Enhancem.: Improvements for -autogui option Bugfix: inst_sge -x -auto ... fails Bugster: 6774221 Issue: 2797 Review: MD RD-2008-11-24-0: Bugfix: corrected complex man page for non-multiplied consumables Review: AH EB-2008-11-24-0: Bugfix: - added missing functions to Perl jsv - fixed bug in master which handled -o switch - fixed qlogin, qsh, qrsh so taht JSV is executed in client environment Review: MD HP-2008-11-21-1: Bugfix: fixed several minor bugs, type casts, wording, added debug output and cleaned things up that appeared while testing on Windows Vista Review: MD HP-2008-11-21-1: Enhancem.: added worker.exe for native Windows to utilbin for testing purposes - not to distribution, yet. Added: worker.c, Makefile Review: MD HP-2008-11-21-0: Bugfix: on Windows, execd installation failed when hostname command returns full qualified name Changed: inst_execd.sh Bugster: 6774847 Issue: 2806 Review: MD EB-2008-11-21-0: Bugfix: Master installation with JMX support fails on DARWIN Enhancem.: mask special characters contained in env when they are passed to JSV scripts Bugster: 6774811 Issue: 2803 Review: MD, JG RD-2008-11-21-0: Enhancem.: analyzed generated events and reduced to minimum where possible. Bugster: 6724000 Issue: 2799 Review: JG LP-2008-11-20-0: Enhancem.: JMX is installed be default, new flag -no-jmx Enhancem.: Improvements for -autogui option Bugfix: inst_sge -x -auto ... fails Bugfix: inst_sge -m -jmx may never finish Bugster: 6774221, 6774228 Issue: 2797, 2798 Review: MD DG-2008-11-20-0: Bugfix: - ibm-loadsensor should be automatically copied to correct location during installation. - typo Bugster: 6315286 Issue: 2796 Review: MD EB-2008-11-20-1: Bugfix: - fixed problem with lost data in bourne shell JSV scripts on platforms were read_raw must be used instead of "read -r" Review: MD JG-2008-11-20-0: Enhancem.: Optionally use jemalloc instead of system malloc on Linux. Review: RD EB-2008-11-20-0: Bugfix: - trigger soft and then hard shutdown of JSVs when master is terminated - fixed core of master in combination of missing user and type on certain archs Review: MD EB-2008-11-19-1: Bugfix: - 2nd job verification is only done if JSV is executed when jobs are added to master - fixed bug introduced with EB-2008-11-19-0 Rreview: MD (RD) RD-2008-11-19-0: Bugfix: - fixed two JSV related memory leaks Review: EB EB-2008-11-19-0: Bugfix: - fixed several minor issues concerning JSV Review: MD DG-2008-11-18-1: Bugfix: Verification of qlogin and qsh does always return error code 0. Bugster: 6726233 Issue: 2731 Review: MP DG-2008-11-18-0: Cleanup: Fixed broken build on irix Review: EB EB-2008-11-17-1: Bugfix: - fixed broken build on irix Review: DG JG-2008-11-17-0: Bugfix: - inst_sge -rst: can't resolve username "none" - inst_sge -rst: doesn't restore resource_quotas and advance reservation with classic spooling Bugster: 6751309, 6751310 Issue: 2791, 2789 Review: MD TB-2008-11-17-0: Bugfix: improved error handling in perl JSV script for list/non-list PARAMs Review: MD EB-2008-11-13-0: Enhancem.: - added JSV URL to submit dialog of qmon and to global configuration - added missing parsing code - fixed bug in parsing code of JSV URLs - added code comments - moved error/info messages to msg files - fixed bug in JSV scripts - disabled logging in JSV scripts Review: MD CR-2008-11-12-0: Bugfix: fixed test_issues module test binary Review: AA TB-2008-11-12-0: Enhancem.: - added Perl implementation for JSV template Bugfix: - fix in TCL implementation for JSV template Review: EB DG-2008-11-12-0: Bugfix: Bugfixes regarding the upgrade procedures. Merged from: V62_BRANCH (DG-2008-10-30-1) Merged from: V62_BRANCH (DG-2008-11-03-0) Merged from: V62_BRANCH (DG-2008-11-06-0) Merged from: V62_BRANCH (DG-2008-11-10-0) Merged from: V62_BRANCH (DG-2008-11-10-1) Bugster: 6765660, 6766830, 6767855, 6767792, 6768967 Issue: 2779, 2778, 2782, 2771 Review: JG, MD, CR JG-2008-11-10-0: Bugfix: fixed a qmaster deadlock on linux fixed minor memory leaks Merged from: V62_BRANCH (JG-2008-10-30-0) Review: RD EB-2008-11-06-0: Enhancem.: - added TCL implementation for JSV template - parsing code for JSV output - functionality to test parsing if JSV related is executed within TS - added man page jsv.1 - source code documentation - fixed several adoc headers which broke generation of pdf files Review: MD JG-2008-11-06-0: Bugfix: When a shadow master took over, the exec daemons do not connect to the new master. Introduced in 6.2u1 development. Review: CR CR-2008-11-06-0: Cleanup: Fixed compiler warnings on hp11 in cl_ssl_framework.c (cl_tcp_framework.c already contains the changes). Belongs to: CR-2008-10-30-1 Review: MD JG-2008-11-04-0: Bugfix: auto install fails with spooling via BDB server Bugster: 6767198 Issue: 2772 Review: MD RD-2008-11-03-0: Enhancem.: added per job consumable functionality to avoid slot multiplication of requests for PE jobs Review: CR, AA (qmon) EB-2008-11-02-0: Enhancem.: first checkin of JSV - functional specification - JSV script template - prepared master for JSV execution - JSV start/stop in clients and master Review: MD NOTE: development is not finished. JSV functionality does not work now. Also documentation is not finished. Final checkin will follow within the next 2 weeks. MD-2008-10-30-0: Bugfix: creation of local config during execd install broken BelongsTo: MP-2008-10-22-1 Review: JG CR-2008-10-30-0: Bugfix: qmaster might be unresponsive for a time because of commlib timeout Merged from: V62_BRANCH (CR-2008-10-23-1) Bugster: 6754568, 6767234 Issue: 2747, 2775 Review: HP o added new module test binaries test_cl_issue_2747_client test_cl_issue_2747_server o enhanced commlib thread list by thread type o qmaster threads now use commlib log list for communication logging o cleanup of commlib internal connection state model o cleanup of commlib functions used code copies o fixed bug for removing connections from connection hash o fixed bug in ssl shutdown behaviour o fixed bug in Makefile of commlib/sheperd (missing dependency for libcomm_static.a) o Fixed possible deadlock in IJS (repairs hanging sheduler problem of CR 6751413) GH-2008-10-29-0: Bugfix: PVM start/stop scripts should not put queue in Error state Bugster: 6727832 Issue: 2714 Review: CR RD-2008-10-28-0: Cleanup: - removed now from internal sconf - replaced array by CULL List - some minor format changes Review: JG RD-2008-10-27-0: Bugfix: fixed scheduler crash with non-consumable RQS Merged from RD-2008-10-22-0 in V62_BRANCH Bugster: 6762505 Issue: 2762 Review: AH MP-2008-10-22-1: Bugfix: qsub -w e does not acknowlegde dynamic load parameters Merged from MP-2008-10-21-0 in V62_BRANCH Bugster: 6554872 Issue: 2240 Review: JG MP-2008-10-22-0: Bugfix: qconf -Aconf modifies existing configuration Merged from MP-2008-10-21-1 in V62_BRANCH Bugster: 6709552 Issue: 2752 Review: JG MD-2008-10-20-0: Bugfix: o SGE_EXECD_PORT gets accepted even when the same as SGE_QMASTER_PORT Bugster: 6750688 Issue: 2754 Review: HP LP-2008-10-17-0: Enhancem.: Introduced -autogui option to inst_sge Bugfix: Cannot choose admin user when installing as root - installation fails Bugfix: Installation might fail if SGE_CELL directory exists Bugfix: Service Tags support displays messages on qmaster startup Bugfix: Upgrade in copy mode should suggest spooling method from the backup first Bugfix: Parallel execd autoinstallation might fail when using only SMF Bugster: 6739725, 6742941, 6743003, 6743012, 6743001 Issue: 2694, 2707, 2710, -, 2709 Review: MD JF-2008-10-16-0: Bugfix: qmod does not follow admin host rules fixed Bugtraq: 6707098 Belongs to: JF-2008-10-06-0 Issue: 2582 Review: JG MP-2008-10-16-0: Bugfix: invalid exit code if host is removed from admin host list during installation Bugster: 6748262 Issue: 2755 Review: DG MD-2008-10-16-0: Bugfix: o spooldefaults core dumps when bootstrap file does not exists Bugster: 6710511 Issue: 2753 Review: JG DG-2008-10-16-0: Bugfix: subordinate properties lost on qmaster restart Bugster: 6319223 Issue: 1741 Belongs to: EB-2008-09-15-0 Review: EB GH-2008-10-10-0: Bugfix: scheduler failure when job-id's roll over to 1 Bugster: 6682558 Issue: 2564 Review: EB MP-2008-10-08-0: Bugfix: SGE_Helper_Service man page still refers to old complex variable when submitting GUI jobs. Also there is a new environment variable that is not yet documented. Bugster: 6637152 Issue: 2751 Review: JG JG-2008-10-08-0: Bugfix: qalter allows a user to place a job in a project forbidden by ACLs Bugster: 6752921 Issue: 2737 Review: CR MP-2008-10-06-0: Bugfix: qhost -l followed by multiple resource/value pairs only accepts the first attribute and ignores the rest. Merged from MP-2008-09-23-0 in V61_BRANCH Bugster: 6734764 Issue: 2684 Review: DG AA-2008-10-06-0: Enhancem.: - added missing jmx function mappings - minor cleanups - jgdish connectjmx enhancement and -ks switch Review: RH, CR RD-2008-10-06-0: Bugfix: - qmaster always assumed a job duration of 0 in verification mode (qalter -w) Enhancem.: - added -w p (poke) option as replacement for old qstat -j Issue: 2750, 2548 Bugster: 6755684, 6685496 Review: JG, CR JF-2008-10-06-0: Bugfix: qmod does not follow admin host rules fixed Bugtraq: 6707098 Issue: 2582 Review: LP RC-2008-10-05-0: Enhancem.: - Refreshed the PPC/PPC64 port - Fixed the SGE-SSH integration link problem Review: pending MP-2008-10-02-0: Bugfix: Modification of submit man page to reflect the correct default behaviour of -w for qrsh, qsh and qlogin Merged from MP-2008-10-02-0 in V61_BRANCH Bugster: 6726233 Issue: 2731 Review: JG JG-2008-10-01-0: Enhancem.: Improved handling of invalid requests in sge_execd. Review: RD DG-2007-10-01-1: Bugfix: Man page issue regarding qresub and the -h n parameter. Issue: 2746 Bugster: 6752920 Review: MP RD-2008-09-30-0: Enhancem.: - Improved scheduling time by using hash Bugfix - runtime limits were delayed after execd restart Issue: 2743 Bugster: 6754094 Review: CR CR-2008-09-29-0: Bugfix: execd reconnect can take up to 90 seconds after qmaster restart Cleanup: cleanup of queue - initial_state implementation Bugster: 6753642 Issue: 2741 Review: RD EB_2008-08-09-29-0: Enhancem.: relaxed accounting - write accounting record if job is deleted with forced attribute. Online usage will be used for accounting - write job logging record if job is deleted with forced attribute - enable master side limit checking when ENABLE_ENFORCE_MASTER_LIMIT is set as qmaster_param (tightly and loose integrated PE jobs are handled correctly). - triggers are added/removed when hosts change into unknown/known state (kill, qconf -ke) - enable mapping of regular qdel to forced qdel if ENABLE_FORCED_QDEL_IF_UNKNOWN is set as qmaster_param - qmaster enforced deleted jobs (h_rt limit) will be logged in accounting file as failed with status 37 ("qmaster enforced h_rt limit") - changed some man pages Bugster: 6702908 Issue: 2551 Review: AH (JG for corresponding changes in V61ARsnapshot3_BRANCH => EB_2008-08-08-08) RD-2008-09-29-1: Bugfix: fixed broken build on irix Review: JG RD-2008-09-29-0: Bugfix: Fixed potential qmaster crash with schedd_job_info and qsub -w jobs Bugster: 6749902 Issue: 2742 Review: JG RD-2008-09-25-0: Bugfix: Fixed issue with overlapping -masterq and -q request Bugster: 6730437 Issue: 2736 Review: DG DG-2008-09-25-0: Bugfix: Fixed return values of -help and "no option" for client tools. Fix of man page regarding -V option. Added help output to stderr when no option is given to the clients. Bugster: 6750325, 6749921 Review: RD JG-2008-09-22-0: Bugfix: in csp mode commlib clients do not respect commlib parameters for the first message Bugster: 6750834 Issue: 2734 Review: CR AP-2008-09-19-0: Task: drmaa junit test Issue2608test.java - output redirection Issue: testsuite issue 180 Review: CR RD-2008-09-19-0: Bugfix: qalter does not print any output and has wrong exit code Belongs to: RD-2008-09-18-0 Review: JG MP-2008-09-18-0: Bugfix: qrsh -w v does not give any feedback to the user after submitting the command. Also, even with no error, the return code is always 1. Merged from V61_BRANCH MP-2008-09-17-0 Bugster: 6726233 Issue: 2731 Review: JG RD-2008-09-18-0: Bugfix: fixed verbosity of qalter -w v Bugster: 6730487 Issue: 2680 Review: CR CR-2008-09-17-0: Bugfix: Commlib does not compare hosts with strcasecmp Issue: 2728 Bugster: 6748854 Enhancem.: Added new module test for simple commlib issues: test_commlib_issue Review: RD CR-2008-09-16-0: Bugfix: execd failed to reconnect to shadow master host Issue: 2700 Bugster: 6738091 Review: RD MP-2008-09-16-0: Bugfix: Parameter PE_RANGE_ALG is shown as SELECT_PE_RANGE_ALG in the man page of sched_conf Bugster: 6567413 Issue: 2723 Review: DG DG-2008-09-16-0: Bugfix: qresub on -b y jobs raises wrong error condition Bugster: 6727295 Issue: 2726 Review: JG EB-2008-09-15-0: Bugfix: subordinate properties lost on qmaster restart Issue: 1741 Bugster: 6319223 DG-2008-09-10-0: Bugfix: Logging into schedd_runlog is never turned off. Bugster: 6744738 Issue: 2717 Review: JG MP-2008-09-04-0: Bugfix: The ibm-loadsensor script has the 5.3.0 case missing in the section that computes the memory load values. Merged from V61_BRANCH MP-2008-09-03-0 Bugster: 6740546 Issue: 2715 Review: AP AP-2008-09-03-0: Bugfix: weight_priority is missing from sched_conf(5) Bugster: 6575735 Issue: 2279 Review: DG DG-2008-08-19-0: Bugfix: qconf -help showed different identifier for the same variable (obj_id_lst and obj_id_list). obj_id_list is consistent with other identifiers. Bugster: 6622128 Issue: 2693 Review: AA MP-2008-08-12-0: Bugfix: qmaster reinstall overwriting an existing installation fails Merged from V61_BRANCH MP-2008-05-29-0 Bugster: 6673742 Issue: 2594 Review: JG, MD AA-2008-08-11-0: Bugfix: sge_ca script fails for installation as root user Bugster: 6735875 Issue: 2685 Review: RH DG-2008-08-08-0: Bugfix: Manpage sge_conf(5) was wrong about default loglevel. Bugster: 6575737 Issue: 2280 Review: JG DG-2008-08-07-1: Cleanup: Update of documentation and comments. Belongs to: DT-2007-11-21-0 Bugster: 6558329 Issue: 2253 Review: MD DG-2008-08-07-0: Bugfix: Man page qdel is misleading with the -f option. Bugster: 6702511 Review: CR MD-2008-07-31-0: CVS: Created V62_TAG ------------------------------------------- JG-2008-07-28-0: Bugfix: Memory leak in sge_execd with qsub -v SGE_* or qsub -V. Bugster: 6729411 Issue: 2672 Review: MD HP-2008-07-23-0: Bugfix: On Windows/Interix, the shepherd didn't exit immediately after the qrsh job finished, it took 60 s to finish Review: JG LP-2008-07-23-0: Bugfix: Fixed inst_sge -help output Review: JG LP-2008-07-22-0: Bugfix: Fixed broken upgrade message Bugfix: inst_sge -post-upd did not upgrade win_hosts Bugfix: Wrong user after connectiong to a win host in the post upgrade procedure Bugfix: tr command has wrong replace arguments in some places Bugfix: Installation fails when another cluster already uses RC scripts on Linux Bugster: 6728406, 6727751 Review: MD, AP RN-2008-07-22-0: Bugfix: qmaster crashes while project deletion (qconf -dprj) in a special case Bugster: 6728262 Issue: 2672 Review: JG MP-2008-07-19-0: Bugfix: Fixed spelling mistakes in the man pages. AH-2008-07-18-0: Bugfix: Fixed error that caused tight_integration_massive test to fail. Bugster: 6710983 Issue: 2603 Review: CR Belongs to: AH-2008-06-10-0 LP-2008-07-18-2: Cleanup: Renamed load_cc.sh and save_cc.sh to load_sge_config.sh and save_sge_config.sh LP-2008-07-18-1: Bugfix: Upgrade did not update N1 windows helper service (Using service name instead of display name now) Bugfix: Wrong users were connecting to windows hosts in the post upgrade procedures Bugfix: Fixed broken -winsvc in certain cases Review: AP AA-2008-07-18-0: Bugfix: o qconf invalid values and exit code fixes Bugster: 6618016, 6618012, 6617973 Issue: 2662 Review: MD MD-2008-07-18-0: Bugfix: o Installation of execution daemon left user unclear which port was chosen. o Wrong permissions if install_qmaster creates qmaster spool directory Bugster: 6726344, 6726285 Issue: 2661, 2660 Review: CR LP-2008-07-18-0: Enhancem: Introduced new inst_sge option -start-all to start complete cluster Enhancem: Improved upgrade messages Enhancem: Upgrade now handles /etc/services sge ports Cleanup: Removed unsed procedure UninstWinHelperPre62 Bugfix: shadow_masters file might contain some hosts twice Bugster: 6675592 Issue: 2521 Review: JG, MD JG-2008-07-17-0: Bugfix: Fixed minor memory leaks. EB-2008-07-17-0: Bugfix: Qmaster message file contains logging about NULL ptr passed to host_list_locate() Bugster: 6726283 Issue: 2658 Review: JG JG-2008-07-14-0: Bugfix: Fixed minor memory leaks. Bugster: 6724877 Issue: 2650 Review: HP EB-2008-07-17-0: Bugfix: Qmaster message file contains logging about NULL ptr passed to host_list_locate() Bugster: 6726283 Issue: 2658 Review: JG AP-2008-07-17-0: Bugfix: Wrong option in upgrade procedure. Review: AS HP-2008-07-16-0: Bugfix: Using fix HP-2008-07-14-0 for all architectures now Cleanup to: HP-2008-07-14-0 Review: CR DG-2008-07-16-0: Bugfix: qsub -w v always returned success. Bugster: 6724949 Issue: 2652 Review: JG MP-2008-07-14-0: Bugfix: Merged from V61_BRANCH: MP-2008-7-11-0 Bugster: 6716082 Issue: 2644 Review: JG JG-2008-07-14-0: Bugfix: Fixed minor memory leaks. Bugster: 6724877 Issue: 2650 Review: EB, HP, CR HP-2008-07-14-0: Bugfix: Fixed wrongly set signals in commlib threads Bugster: 6725446 Issue: 2653 Review: EB, CR MD-2008-07-14-0: Bugfix: /etc/services is not handled correctly during installation Bugster: 6711517 Issue: 2655 Review: JG RN-2008-07-14-0: Bugfix: 'Infinity' must be rejected when specified in 'complex_values' or RQS limits for consumables Bugster: 6613438 Issue: 2391 Review: AH CR-2008-07-11-0: Bugfix: - Merged from V61_BRANCH: CR-2008-07-09-0 - Execd Cleanup Review: MD HP-2008-07-11-0: Bugfix: Added workaround to set missing resource limits Bugster: 6722507, 6533487 Issue: 2370 Review: AS, MD JG-2008-07-11-0: Bugfix: Fixed minor memory leaks. Bugster: 6724877 Issue: 2650 Review: MD JA-2008-07-11-1: Bugfix: - improve precondition in sge_task_depend_get_range to accept any valid task id - remove warnings and debug messages associated with sge_task_depend_get_range failure Review: AH,JG JA-2008-07-11-0: Bugfix: Tasks held with array dependency may get deleted prematurely Bugster: 6723951 Issue: 2628 Review: AH AA-2008-07-10-0: Bugfix: - install_qmaster -jmx fails if non default keystore path is used - jmx keystore property is wrong for non default keystore location - use LinkedHashMap to keep sort order - jgdi cleanups - GridCA -days extension Bugster: 6723932, 6724576 Issue: 2647, 2646 Review: RH, CR LP-2008-07-10-0: Cleanup: Removed inst_update.sh Bugfix: Fixed typos & usages Review: MD JG-2008-07-10-0: Bugfix: Error message of qalter -q "" suggests memory corruption. Bugster: 6704393 Issue: 2575 Review: MD ZG-2008-07-10-0: Bugfix: Default is the local spool directory when install_execd is running on a Windows host Bugfix: Improved text at "Execd spool directory configuration" at execd installation Bugster: 6589820 Issue: 2634 Review: LP LP-2008-07-09-0: Enhancem: Improved upgrade mode detection Cleanup: Renamed LP-2008-07-07 to LP-2008-07-07-0 Bugfix: ServiceAlreadyExists error on non-smf hosts Bugfix: Fixed upgrade remote actions Bugfix: Upgrade uses same local execd spool dir in copy mode Bugster: 6723936 Review: JO JG-2008-07-08-0: Bugfix: qmaster runs out of memory on AIX Bugster: 6720078 Issue: 2641 Review: MP MP-2008-07-08-0: Enhancem: Updated the test_flatfile_spool file to take into account parsing of RQSes with a backslash followed by a line break combination Review: RD RD-2008-07-08-1: Cleanup: - removed references to DRMAA 0.95 from man pages - deprecated obsolete parameters/switches - removed redundant upgrade documents Review: AA RD-2008-07-08-0: Cleanup: - fixed and enhanced test_performance module test - some smaller code cleanups Review: AA LP-2008-07-07-0: Enhancem: Removed ST question, ST used by default Enhancem: ST also register PRODUCT_PARENT_URN Bugfix: execd install fails on OS X when qmaster boot scripts are present Bugfix: upgrade does not save and restore sge_qstat, sge_ar_request,sge_qquota Bugfix: install fails with mutually dependent subordinate queues Bugster: 6716772, 6722920, 6723130 Issue: 2574, 2625 Review: RD, ZG MP-2008-07-07-0: Bugfix: RQS: Line wrap of host list introduces syntax error Bugster: 6647676 Issue: 2449 Review: DT RD-2008-07-07-0: Bugfix: fixed broken auto installation Review: AA DG-2008-07-07-0: Bugfix: Java DRMAA final static variables are accessed through object references. Bugster: 6722930 Review: JG LP-2008-07-04-0: Enhancem: Windows host list is now saved during the backup for upgrade procedure Enhancem: SGE_CELL, SGE_CLUSTER_NAME, PRODUCT_MODE is used from the backup during the upgrade Enhancem: Added post update actions to clean execd_spool_dirs, remove RC scripts, create RC scripts, update WINSVC Enhancem: WINSVC install no longer reinstalls a service (winupdate should be used) Cleanup: Various upgrade/install code Bugfix: JMX java detection fixed Bugfix: Upgrade script failed when choosing a different port number then $SGE_QMASTER_PORT Bugfix: upgrade does not create arusers user set Bugfix: upgrade does not save host_aliases file Bugfix: upgrade does not create display_win_gui complex Bugfix: Auto install template value checking is too strict on COPY_COMMAND Bugfix: qmaster installation displays "No CSP system installed" twice when no CSP Bugster: 6717816, 6722652, 6722805, 6722806, 6717891, 6722925 Review: AP DG-2008-07-04-0: Cleanup: Java DRMAA Junit test: problems with build numbers fixed. Merged from 61_BRANCH DG-2008-07-03-0 Review: CR JG-2008-07-04-0: Bugfix: Same path for input and output/error file must be rejected. Bugster: 6310194 Issue: 1751 Review: RD JG-2008-07-03-1: Enhancem.: New default location for Berkeley DB and openssl. Review: AA JG-2008-07-03-0: Enhancem.: Updated review template and help document. Review: sge-et AH-2008-07-03-1: Bugfix: Invalid qconf -mrqs crashes qmaster with segmentation fault Bugster: 6722217 Issue: 2432 Review: TH,RD MP-2008-07-02-0: Bugfix: qstat -j does not print array task information Bugster: 6710485 Issue: 2606 Review: JG HP-2008-07-02-0: Bugfix: shepherd wasn't able to read user certificates Bugster: 6718433 Bugfix: insufficient handling of commlib errors Bugster: 6718447 Bugfix: incorrect accounting when terminate method is used Bugster: 6718453 Bugfix: qrsh and qlogin has very slow response to user input Bugster: 6686921 Review: JG DG-2008-06-30-0: Cleanup: Missing depend target in qmake makefile. Review: AA ZG-2008-06-25-0: Bugfix: incorrect dependency on xinetd in init scripts fo linux Bugster: 6677192 Issue: 2523 Review: JG ZG-2008-06-23-0: Bugfix: install_execd -winsvc should output wrong usage Bugster: 6692104 Issue: 2624 Review: JG DG-2008-06-23-0: Cleanup: Temporary load file is not deleted by script. Cleanup of check in EB-2008-04-16-0. Review: JG RH-2008-06-19-0: Bugfix: copy paste error in jgdish xmldump fixed Review: AA AA-2008-06-19-0: Bugfix: sge_ca script does not allow easy sdm_admin keystore creation Bugster: 6716648 Issue: 2619 Review: RH DG-2008-06-19-0: Cleanup: Outdated installation hint for AIX loadsensor in script comments. Review: EB JG-2008-06-16-0: Bugfix: reporting file is lacking information about global consumables, if log_consumables=false Bugster: 6710542 Issue: 2615 Review: HP AA-2008-06-16-0: Bugfix: - certtool.sh didn't overwrite key dir for reinstall in testsuite - NullPointerException fix for sharetree Review: JG LP-2008-06-13-1: Bugfix: Fixed SMF specification document LP-2008-06-13-0: Enhancem: Added SMF specification document DG-2008-06-13-0: Bugfix: job hold due to -hold_jid is not indicated as STATE_SYSTEM_ON_HOLD by drmaa_job_ps(3) Issue: 2608 Bugster: 6562277 Review: AH, RD AH-2008-06-12-0 Bugfix: - qalter -w v is silent on finished jobs - qalter -w v wrongly changes job attributes Bugster: 6708891,6712685 Issue: 2598,2607 Review: RD RD-2008-06-12-1: Bugfix: fixed possible segfault introduced with RD-2008-06-12-0 Review: DG RD-2008-06-12-0: Bugfix: load_sensor does not work with reports for other hosts Bugster: 6713785 Issue: 2611 Review: DG RD-2008-06-11-0: Bugfix: mtmalloc on Solaris was wrongly disabled Review: JG AH-2008-06-10-0: Bugfix: qsub option -q breaks -masterq Bugster: 6710983 Issue: 2603 Review: AA RD-2008-06-10-0: CVS: Created V62beta2_TAG ------------------------------------------------ LP-2008-06-09-0: Bugfix: Upgrade load order fixed Review: RD AA-2008-06-06-0: Bugfix: load scaling display not working correctly sensitivity of buttons in Queue Control dialog Bugster: 6711578 Issue: 2605 Review: AH LP-2008-06-06-0: Enhancem: Various upgrade procedure enhancements and fixes Review: RD RD-2008-06-06-0: Bugfix: - installation fails with ubuntu 08.04 Issue: 2604 Review: JG RD-2008-06-05-1: Bugfix: - added support for lx26-ppc64 - imported several compiler error fixes from community Issue: 2584, 2585, 2348, 2587, 2588, 2590 Review: RD-2008-06-05-0: Bugfix: Service Tags version is incorrect Bugster: 6698169 Review: JG RD-2008-06-04-1: Bugfix: corrected max_advance_reservation entry in sge_conf(5) Review: AH LP-2008-06-04-0: Enhancem: New upgrade 6.0+ to 6.2, new method introduced to upgrade to different SGE_ROOT, SGE_CELL and tranfer old configuration to a new cluster Bugster: 6447131 Bugfix: install fails creating unique cluster name Bugster: 6709231 Review: RD, PJ, AP EB-2008-06-04-0: Bugfix: schedd_runlog file is not written when qconf -tsm is used Review: JG Bugster: 6710417 Issue: 2602 RD-2008-06-04-0: Bugfix: added fix for deleting jobs with no enrolled an no unenrolled tasks Review: JG DG-2008-06-03-0: Bugfix: Added missing space character in inst_st.sh Review: AH PJ-2008-06-02-0: Bugfix: service tags cannot be disabled from auto-install Bugster: 6709229 Issue: 2600 Review: LP RD-2008-06-02-0: Cleanup: minor AR related changes Review: AA CR-2008-06-02-1: Cleanup: Fixed message macros Review: RD AA-2008-06-02-0: Bugfix: - qrsub usage for -now y/n Bugster: 6707944 Review: RD CR-2008-06-02-0: Cleanup: commlib: - cl_com_cached_gethostbyaddr(): more error tolerant compare of reverse ip resolving - report ip resolve errors to application utilbin/gethostbyaddr: - added application error callback for commlib Bugster: 6715366 Issue: 2616 Review: RD AA-2008-05-30-0: Bugfix: - upgrade to openssl-0.9.8g - qmon AR fixes - make libjgdi.so compile on AIX and Irix Bugster: 6706406, 6715346 Review: CR JG-2008-05-30-0: Bugfix: 32-bit Linux libraries are having problems with file access in 64-bit NFS environments Bugster: 6707055 Issue: 2543 Review: AH AH-2008-05-30-0: Bugfix: sge_schedd can core dump when resoruce quotas are used Bugster: 6707847 Issue: 2546 Review: JG RN-2008-05-29-3: Bugfix: use of -l tmpdir=abc can crash schedd Bugster: 6650497 Issue: 2597 Review: AH CR-2008-05-29-0: Bugfix: Communication library cleanup: o Fixed wrong lease time information for ingoing messages for qping -dump output. Lease time for ingoing messages is not available because dump message is not generated when message is polled by application. o Fixed broken qping -dump output for connect message o CM protocol message rollback: CM message now also contains message destination endpoint name which was removed by performance changes. The information is needed in service to do hostname resolve and component id checks. Bugster: 6715371, 6715373 Issue: 2617, 2618 Review: AA RN-2008-05-29-2: Bugfix: o qconf -ae|-Ae return 0 even if exechost exists already o qconf -dxxxx does not set exit status on error o qconf -as, -am, -ao, -Ae, -Acal, -Ackpt, -Ap when msg "already exist" should return not 0 exit code o qconf -acal doesn't return error code 1 when failed Bugster: 6652411,6546280,6704283,6597558 Issue: 2595 Review: JG RN-2008-05-29-1: Bugfix: Incorrectly considering two host group names to be the same Bugster: 6666161 Issue: 2567 Review: JG RN-2008-05-29-0: Bugfix: Failed to deliver STOP signal for subordinated jobs Bugster: 6623174 Issue: 2478 Review: JG JO-2008-05-28-0: Bugfix: o Argument not passed from SetupRcScriptNames() Bugster: 6715369 o Corrected arco_mysql.txt Review: LP HP-2008-05-27-0: Enhancem: Changed name of Windows helper service from N1 Grid Engine Helper Service to Sun Grid Engine Helper Service HP-2008-05-23-0: Enhancem: Moved Windows source to gridengine repository EB-2008-05-21-0: Doc: Changed some man pages Bugster: 6680939, 6659269, 6653487, 6660407, 6638371 Issue: 2577, 2578, 2579, 2484, 2580 Review: DG HP-2008-05-19-0: Bugfix: New interactive job support didn't work properly in secured mode. Enhancem: Added infos about new interactive job support to man pages Review: AA Remark: This bugfix was not sufficient, a complete bugfix will follow! AH-2008-05-19-0: Bugfix: Incorrect documentation for xprojects setting in sge_conf(5). Bugster: 6704271 Issue: 2573 Review: RD DG-2008-05-15-0: Enhancem.: HP-UX 11 64 bit machines now also compiling the HP-UX 11 32 bit sources. Review: JG MP-2008-05-15-0: Bugfix: SGE util/arch script is broken for AIX 5.3 os Bugster: 6648160 Issue: 2572 Review: JG DG-2008-05-14-0: Cleanup: aimk clean didn't clean 32 bit libraries when called on a 64 bit Solaris machine. Review: AA DG-2008-05-07-0: Cleanup: Copyright file handling didn't work in Netbeans IDE since env.AIMK_COPYRIGHT_DIR is not set Review: AA MD-2008-05-07-0: Bugfix: o setting of QMaster port number leads to infinite loop Bugster: 6696331 Issue: 2569 Review: HP AA-2008-05-06-0: Bugfix: adds unexpected files fix Review: DG RD-2008-05-05-2: CVS: Created V62beta ------------------------------------------- RD-2008-05-05-1: Cleanup: - added missing copyright notice Review: AA RD-2008-05-05-0: Cleanup: - added copyright files to jar files - removed Author from DRMAA JavaDoc Review: AA EB-2008-05-05-0: Doc: removed sge_schedd man page changed several man pages Review: DG RD-2008-05-02-0: Doc: corrected copyright year and fixed references to documentation Review: JG HP-2008-05-02-0: Bugfix: Use of commlib debugging feature caused deadlock on lx26-amd64 Changed: common/sge_ijs_comm.c AH-2008-04-30-0: Bugfix: - use libcore.so to fix problem with no core files when SGE daemons crash under Linux - fixed install dialogue wording that became necessary after to AH-2008-04-10-0 Bugster: 6686655 Issue: 2552 Review: RD RD-2008-04-30-1: Cleanup: Removed unused QU_queue_number Review: EB EB-2008-04-30-1: Cleanup: Removed late lock; cleanup man pages Review: RD RD-2008-04-30-0: Enhancem.: disabled execution of JC_FILTER code if feature is disabled Review: EB HP-2008-04-30-1: Bugfix: fixed several small bugs reported by insure Bugfix: now rsh_daemon config is obeyed only for qrsh without command, qlogin_daemon config only for qlogin, and so on Bugfix: partial bugfix of SHLIB_PATH problem on HP1164 Bugfix: -pty yes|no switch affects setting of terminal mode Bugfix: if terminal mode can't be set, now qrsh prints error and exits Review: JG RD-2008-04-29-1: Bugfix: fixed memory leak introduced by RD-2008-04-28-0 Review: JG RD-2008-04-29-0: Enhancem.: added AR_submission_time to accounting and job reporting Review: JG JG-2008-04-29-0: Bugfix: fixed warning "attempting to commit an event master transaction, but no transaction is open" when submitting interactive jobs (-now yes) Bugster: 6690757 Review: RD AA-2008-04-29-0: Cleanup: - JMX ssl set as default - default value NONE for examples - corrections in examples - correction of comments in management.properties Review: RH LP-2008-04-29-1: Bugfix: sgemaster does not start/stop both qmaster, shadowd SMF services Bugster: 6695278 Review: JG LP-2008-04-29-0: Bugfix: Stopping qmaster may stop shadowd SMF service in some cases Bugfix: sge_smh.sh does not work as standalone script Bugster: 6692677, 6692668 Review: JG MD-2008-04-28-0: Bugfix: o Set SGE_QMASTER_PORT in settings file if sge_qmaster is not found in /etc/services file (bug was not really fixed with checkin: OJ-2008-01-31-0), added a dialog and changed existing. Bugster: 6589297 Issue: 2477 Review: HP, DG Changed: installation RD-2008-04-28-0: Enhancem. - reduced queue element send by job start - added BDB transactions for job script and job Cleanup: - removed last occurences of QSTD - removed MAX_ORDER_LIMIT Review: EB AA-2008-04-24-0: Cleanup: - renamed java drmaa directory to libs/jdrmaa and added nbproject files Review: DT, DG RD-2008-04-22-0: Bugfix: - ar_id is not contained in qstat -j output - AR submit time is missing in some reporting lines Bugster: 6691984, 6691975 Review: JO RD-2008-04-21-1: CVS: Created V62beta_RC2_TAG --------------------------------------------------- HP-2008-04-21-0: Enhancem: added -pty switch to qrsh Enhancem: several cleanups Changed: shepherd, qrsh Review: EB AA-2008-04-21-1: Bugfix: - suppress pw echo for JMX - enable pw in JGDIShell Review: RH AA-2008-04-21-0: Bugfix: - javadoc comment fixes - darwin missing function fix - distinst OPENSSLSOVERSION fix - aimk.site lx24-ia64 java support fix Review: CR RD-2008-04-21-0: Enhancem.: - removed SIMULATE_EXECDS param - enhanced SIMULATE_JOBS Review: JG JG-2008-04-21-0: Bugfix: Deleting user defined complex variables failed, builtin host complex variables could be deleted. Review: MD DG-2008-04-18-0: Cleanup: Path to tar file was hardcoded. Wasn't on all architectures in the same path. Review: AH LP-2008-04-18-0: Bugfix: inst_sge -ux fails for root user on a filesystem mapped with no-root permissions Bugfix: installation and autoinstallation does not work on SMF supporting hosts Bugster: 6687514, 6685565 Bugfix: Fixed drmma test testGetVersion Review: JG RD-2008-04-18-0: Bugfix: fixed qmaster-scheduler sync issue which led scheduler schedule a job twice Bugster: 6690182 Review: JG CR-2008-04-17-0: Cleanup: o make use of hash table switchable for commlib lists o make cl_com_ignore_timeouts() signal handler save o cl_connection_list_destroy_connections_to_close(): - fixed wrong/missing locking - cleanup Review: RD JG-2008-04-17-0: Bugfix: reactivated profiling in event_mirror minor fixes and cleanup in scheduler thread mainloop Review: RD EB-2008-04-16-0: Bugfix: o loadsensor does not work on AIX Bugster: 6661060, 6717172 Issue: 2508, 2622 Review: RD DG-2008-04-16-0: Cleanup: Compiler error (indicated argument is incompatible with corresponding formal parameter (NULL was not casted in pointer type) while executing aimk -only-depend on baumbart (IRIX 6.5) Review: JG JG-2008-04-16-0: Enhancem.: Allow insure instrumentation for all platforms. Review: DG JG-2008-04-16-0: Cleanup: fixed typo, cleanup Review: MD AH-2008-04-16-0: Enhancem.: Improvements with cluster queue matching * added consideration for queue resource limits * added consideration for consumables * disregard non-string resources for parallel jobs that request more than a single slot Review: RD JG-2008-04-15-0: Bugfix: loadcheck reported "kstat finds too many cpus" Bugster: 6688666 Issue: 2555 Review: RD RD-2008-04-15-0: Enhancem: - reduced list copy in scheduler Bugfix: - reduced load reports are broken Bugster: 6688549 Review: CR, JG EB-2008-04-15-0: Bugfix: o print worker queue length in message file and qping output o removed second worker queue (test code for performance tests) this code is not needed anymore Review: MD AA-2008-04-14-0: Bugfix: min/max typo leads to wrong qstat output Review: RD RH-2008-04-14-0: Bugfix: Avoid trashing of simulated load values in simulated_execds mode Review: RD RD-2008-04-14-0: Enhancem.: Added messages lost detection for test_virtual_gdi_client Review: CR DG-2008-04-14-0: Cleanup: fix of broken build: -errwarn=%all switched off for special targets (X11, remote) sge_smf compiler warning fix due to errwarn=%all Review: AA RH-2008-04-11-0: Enhancem.: Simulated execds can now get a copy of a load report of a real execd (complex load_report_host must be defined and the simulated execd must have a complex load_report_host=) Review: RD EB-2008-04-11-0: Bugfix: o disabled compiler optimization for Solaris x86 which causes SEGV when threads are shut down o fixed some compiler wanings o now on all Solaris hosts comipler will stop for warnings Review: JG AH-2008-04-10-0: Bugfix: Set schedd_job_info to false in all scheduler profiles to mitigate the problem of schedd_job_info causing immense memory consumption Bugster: 6686621 Issue: 2464 Review: JG CR-2008-04-09-0: Bugfix: commlib might crash if running out of memory Bugster: 6686155 Issue: 2546 Review: MD RD-2008-04-09-0: Enhancem: added hash to scheduler message list Review: JG JG-2008-04-09-0: Bugfix: Fixed non critical memory leaks. Review: EB MD-2008-04-08-0: Bugfix: o Configuration file check of automatic installation does not recognize remsh Bugster: 6673782 Issue: 2549 Review: JG JO-2008-04-07-0: Bugfix: Intermediate acct records are not mentioned in man reporting Bugster: 6572196 Review: JG RD-2008-04-07-0: Bugfix: fixed execd installation with default domain Bugster: 6684415 Issue: Review: CR RD-2008-04-03-2: CVS: Created V62beta_RC1_TAG --------------------------------------------------- LP-2008-04-03-0: Bugfix: Fixed deadlock in execd sig. handler with SMF Review: RD EB-2008-04-03-0: Bugfix: o added allinfo to default qping -info output (did missing review for EB-2008-03-25-0) Review: JG RD-2008-04-03-1: Cleanup: added some code cleanup and smaller performance improvements Review: JG RD-2008-04-03-0: Enhancem: Solaris binaries are now linked by default with mtmalloc Review: JG AH-2008-03-31-0: Bugfix: qsub(1) man page mentioned non extisting file. Issue: 2535 Review: TBD MP-2008-03-27-0: Bugfix: qhost -xml has wrong namespace. Merged from V61 branch Bugster: 6680160 Issue: 2515 Review: MD AA-2008-03-27-0: Bugfix: - rmi codebase missing juti.jar - type in management.properties.template Review: CR RN-2008-03-25-0: Bugfix: o Missing array job task usage in the accounting file o qhost/qstat can't be interrupted with ctrl-c Bugster: 6625420, 6619046 Issue: 2479, 2480 Review: JG MP-2008-03-25-1: Bugfix: Merge of MP-2008-01-10-0 from V61_BRANCH. Bugster: 6517626 Issue: 2455 Review: JG MP-2008-03-25-0: Bugfix: Merge of MP-2008-02-11-0 from V61_BRANCH. Bugster: 6650827 Issue: 2485 Review: MD EB-2008-03-25-0: Enhancem.: - enhanced qping -info monitoring output Review: MD, review for ext_sch_output() still needs to be done GH-2008-03-23-0: Bugfix: Reprioritize disappears after sge_qmaster restart Bugster: 6574269 Review: AH MP-2008-03-21-1: Bugfix: Merge of MP-2008-02-22-0 from V61_BRANCH. Bugster: 6575740 Issue: 2287 Review: JG MP-2008-03-21-0: Bugfix: Merge of MP-2008-02-18-0 from V61_BRANCH. Bugster: 6561179 Issue: 2494 Review: JG, OJ LP-2008-03-20-0: Bugfix: svcadm restart does not restart qmaster/shadowd services Bugfix: SMF execd does not corretly shutdown on kill -15 Bugfix: SMF execd might not go to disabled state if sge admin user does not have smf auths Bugster: 6677918, 6675579, 6675693 Review: HP, PJ HP-2008-03-20-0: Bugfix: o Cleaned up link dependencies of static SSL libraries Now static SSL libraries are only linked to binaries when it's really necessary. All other binaries dynamically load them using dlopen(). Bugster: 6672719 Review: AA MP-2008-03-20-0: Bugfix: Merge of MP-2008-01-31-0 from V61_BRANCH. Bugster: 6629016 Issue: 2467 Review: MD HP-2008-03-19-0: Bugfix: SGE_ARCH variable was not set but used in certtool.sh Review: MD JG-2008-03-19-0: Enhancem. Further event master cleanup Review: EB CR-2008-03-19-0: Bugfix: - added missing sgeE_EXECHOST_MOD event generation for execd hosts in unknown state in sge_load_value_cleanup_handler() Review: JG AA-2008-03-18-0: Bugfix: - sge_ca man page - qping man page typos - qmon help file - qmon restore of settings - mk_ascman on linux option Bugster: 6618332, 6629016, 6660953, 6660954 Issues: 2467, 2517, 2518, 2519 Review: EB JG-2008-03-18-0: Enhancem.: Use thread local storage for transactions in event master. Review: EB JO-2008-03-18-0: Enhancem.: changed detection message Review: LP EB-2008-03-18-0: Enhancem.: changes for darwin-x86 (Leopard) Review: AA CR-2008-03-17-0: Bugfix: Uninstall execd ignored because of hostname matching problem Enhancem.: Log hostname resolve changes into auto install log for (un)installing execd hosts Issue: 2516 Bugster: 6676063 Review: MD LP-2008-03-17-1: Bugfix: shadow_masters file might contain some hosts twice Bugfix: sgemaster -qmaster stop shutdowns also shadowd Bugfix: execd uninstallation should remove execd_spool_dir if it contains only 1 execd (local) Bugfix: Service in maintanence is removed in SMFUnregister() Bugster: 6675592, 6673706, 6675597 Review: PJ LP-2008-03-17-0: Bugfix: SMF services => standalone checkprog crash Bugster: 6675604 Review: PJ JG-2008-03-17-1: Cleanup: Event delivery time for scheduler (schedule_interval) was set twice when schedd config changed. Review: EB JG-2008-03-17-0: Bugfix: Fixed incorrect handling of flush_submit_secs and flush_finish_secs in scheduler thread. Review: EB JG-2008-03-14-0: Bugfix: Fixed possible libdrmaa.so crash introduced with JG-2008-03-13-0. Review: AH AH-2008-03-13-0: Bugfix: Fix possible libdrmaa.so crash in sge_uid2user() Review: DG RD-2008-03-13-0: Bugfix: fixed issue with queue subordinate suspention introduced by RD-2008-03-04-0 Review: AH JG-2008-03-13-0: Bugfix: - fixed minor memory leaks. - specifying more than one drmaa_v_env attribute causes spurious error msg Bugster: 6614108 Issue: 2392 Review: RD AA-2008-03-13-0: Bugfix: changed installation for JMX install_qmaster -jmx instead of enable jmx dialog correction of build.xml for clean target Review: RH RD-2008-03-12-0: Bugfix: - removed wrong unique hash introduced with RD-2008-03-10-0 - replaced loop in qstat by hashed access Review: JG MP-2008-03-11-0: Bugfix: Merge of MP-2007-12-05-0 from V61_BRANCH. Bugster: 2150515 Issue: 2315 Review: JG RD-2008-03-10-0: Enhancem. replaced some loops by hashed access Review: JG HP-2008-03-07-0: Bugfix: qrsh -nostdin failed when it was started with closed stdin Changed: commlib, qsh Bugster: 6670917 Issue: 1258 Review: AA, CR AA-2008-03-06-0: Bugfix: - missing setDefault* functions in ClusterQuueue(Impl) - sge_ca script corrections - jgdi crash fix due to logging Changed: jgdi, sge_ca Review: RH MD-2008-03-06-0: Bugfix: o Changed some script output to increase the usability. Text was hard to understand. o Some typos Changed: installation Review: JG AH-2008-03-06-0: Bugfix: Access_list(5) man page not precise enough with regards secondary/primary group(s) Issue: 2475 Bugster: 6671963 Review: JG AH-2008-03-05-0: Enhancem: Merged code for scheduler matching statistics profiling from DIST6dot1AR branch. Review: AA MP-2008-03-04-0: Bugfix: Fixed error introduced into build through checkin of MP-2008-02-22-0 Bugster: 6647620, 6640688 Review: MD RD-2008-03-04-1: Bugfix: loadsensor does not work or AIX51 Issue: 2508 Bugster: 6661060 Review: AH AH-2008-03-04-1: Docs: Added documentation for qmaster_params ENABLE_RESCHEDULE_KILL and ENABLE_RESCHEDULE_SLAVE to sge_conf(5). Issue: 2491 Bugster: 6670879 Review: RD LP-2008-03-04-0: Bugfix: Uninstallation of remote execd if not interactive Bugfix: Infotext spawned on remote machine with -wait or -ask does not display the text Bugster: 6670352,6669655 Bugfix: Improved uninstallation dialogs Review: PJ RD-2008-03-04-0: Enhancem: - improved qmaster startup time - reduced mem/cpu usage for AR submission - reduced mem/cpu usage for job -w v verification Bugster: 6653424 Issue: 2483 Review: AH AH-2008-03-04-0: Enhancem: Consider also user_list/xuser_list in static cluster queue matching Review: RD EB-2008-03-03-0: Bugfix: added code for additional performance test. disabled logging in locking module. Review: TBD AH-2008-03-03-0: Bugfix Fix drmaa_job_ps() memory leak Issue: 2505 Review: RD JG-2008-02-29-0: Bugfix: Fixed memory leaks in japi, drmaa, test_drmaa. Review: EB JO-2008-02-29-0: Doc: howto arco_mysql.txt - added InnoDB storage engine requirement LP-2008-02-28-0: Bugfix: Uninstall does not remove the SGE_STARTUP_SCRIPT Bugfix: qmaster crashes when SGE_ND=1, dl 2 and BDB server spooling Bugster: 6659880, 6666732 Bugfix: Replaced N1 with Sun in the logs Enhancem: Old cluster_name checked only for root installation Review: PJ PJ-2008-02-27-0: Bugfix: - Wrong detection of sge_qmaster port Bugster: 6659854 Review: LP AA-2008-02-27-0: Bugfix: - default exit func (commlib error: got read error (closing "xyz/qconf/2")) - qmon advanced reservation delete bug - jvm restart - jgdi event test fix Bugster: 6660955 Review: JG, RH JG-2008-02-27-1: Bugfix: Internal event clients didn't have hash tables on their master lists. Review: EB JG-2008-02-27-0: Enhancem.: Event master simplification, cleanup, fixes. First step of bigger cleanup effort. Review: EB RD-2008-02-26-0: Enhancem: - added qmaster_params ENABLE_RESCHEDULE_KILL and ENABLE_RESCHEDULE_SLAVE - reduced amount of data for tight slave tasks - reduced job_usage data in sharetree scheduling orders - improved advance reservation error message - added test for flatfile spooling - flatfile spooling cleanup Issue: 2491 Bugster: 6670879 Review: EB HP-2008-02-25-0: Enhancem.: Added automounter prefix of Darwin to sge_aliases file Changed: dist/util/sge_aliases CR-2008-02-25-0: Bugfix: Auto execd installation did not support adding submit hosts Bugster: 6667218 Issue: 2504 Review: MD EB-2008-02-25-0: Bugfix: removed not needed mutex locking in qmaster Review: MD EB-2008_02-25-0: Bugfix: fixed very critical bug which might cause two or more threads to get the global or confguration lock. Review: CR CR-2008-02-22-0: Bugfix: Disabled delayed commlib listen() for qmaster. Introduced with CR-2008-02-21-0 Some components rely on early qmaster commlib availability (shadowd, qmaster startup on migrate host, etc.) MP-2008-02-22-0: Bugfix: qstat -j output missing fields Bugster: 6657615 Issue: 2501 Review: JG CR-2008-02-21-0: Enhancem.: Enable qmaster connections after complete qmaster startup Bugster: 6665939 Issue: 2500 Review: RD HP-2008-02-21-0: Bugfix: o Fixed broken terminal environment after qrsh ran. o Fixed dropping of last lines of output of a # qrsh cat largefile.txt Bugster: 6662700, 6664827 Changed: qrsh, shepherd, commlib Review: RD, CR AH-2008-02-20-0: Bugfix: Memory leak in drmaa_run_job()/drmaa_run_bulk_job() Bugster: 6665417 Issue: 2497 Review: RD AH-2008-02-19-0: Bugfix: Fix broken aimk -no-secure compilation Review: TBD JG-2008-02-19-0: Bugfix: qsub does not accept resource strings > 256 bytes Bugster: 6664567 Issue: 2495 Review: RD LP-2008-02-19-0: Enhancem.: Fixed typo Bugfix: Auto installation fails with exit state 2 Bugster: 6662616 Issue: 2496 Review: RH RD-2008-02-18-0: Bugfix: - fixed missing projects/users after qmaster restart with sharetree - fixed qmaster crash with long host eval expression - reintegrated slave task job delivery caching lost due to RD-2007-12-21-0 Bugster: 6664502, 6664503 Issue: 2490, 2492 Review: JG EB-2008-02-18-0: Enhancem.: - execd reports will now be handled in worker JG JG-2008-02-18-0: Enhancem.: output event timestamp instead of current time in qevent client Review: RD AH-2008-02-18-0: Bugfix: JAT_granted_pe was not set back to NULL if the advance reservation does not exist at job delivery time Review: TBD EB-2008-02-17-0: Enhancem.: - GDI answer might be send in worker thread instead of listener thread (depends on define SEND_ANSWER_IN_LISTENER) - if answer is send in worker it might be possible to block the listener if it should wait till GDI package is handled (depends on define BLOCK_LISTENER) Review: RD CR-2008-02-15-0: Bugfix: check if local spool directory already was deleted by global spool directory deletion in inst_execd_unist.sh (Additional fix to CR-2008-02-14-0) Review: RD EB-2008-02-15-0: Bugfix: - changed initial number of worker/listerner threads to 2 according to performance tests - added/removed comments Changed: install script Review: RD CR-2008-02-14-0: Bugfix: fixed deletion of execd base spooling directory in inst_execd_uninst.sh Review: RD RH-2008-02-14-0: Bugfix: GridCA500Name was using old grm_daemon constant Changed: juti Review: AA OJ-2008-02-13-0: Bugfix: sgeexecd startup script shouldn't suppress error messages from sge_execd binary Bugster: 6609754 Issue: NA Review: RD, HP Bugfix: Modified common/msg_common.h message to work with IZ 1751 Bugster: 6310194 Issue: 1751 Review: AP AA-2008-02-13-0: Bugfix.: - jgdi client shall use SGE Daemon keystore for JMX ssl mode - jmxeventmon support for SSL, eventmon script - memory leaks qmon, cull Changed: jgdi, install scripts Review: RH RH-2008-02-13-0: Bugfix.: jgdi client could only connect to one qmaster if jmx is running in ssl mode Changed: jgdi Review: AA RD-2008-02-12-0: Bugfix: auto install did not set local execd spool dir Review: CR RH-2008-02-12-0: Bugfix.: jgdi ssl uses wrong client socket factory Changed: jgdi Review: CR RH-2008-02-12-0: Bugfix.: jgdi client ssl setup added Changed: jgdi Review: CR PJ-2008-02-11-0: Bugfix.: Bdb jobscript spooling, job scripts will stay in the database Changed: qmaster Bugster: 6660859 Review: LP HP-2008-02-11-1: Bugfix: qrsh had 1 s delays from time to time Changed: shepherd Review: CR HP-2008-02-11-0: Bugfix: qrsh doesn't work if system runs as unprivileged user Changed: shepherd Bugster: 6660886 Review: CR EB-2008-02-11-0: Bugfix: Improved waiting section at the end of the scheduler main loop Review: CR EB-2008-02-01-0: Bugfix: - sconf_* functions in scheduler access data sent via event mirror now Review: RD LP-2008-02-08-0: Bugfix: Permission problems writing the cluster_name file Bugfix: Unable to remove the SMF startup script Bugfix: missing clear in inst_sge -db Bugfix: Uninstall does not remove the SGE_STARTUP_SCRIPT Bugfix: inst_sge -ux all -um fails Bugster: 6659196, 6659203, 6659195, 6659880, 6660899 Bugfix: Other minor install fixes Review: PJ, RD RH-2008-02-08-0: Bugfix: jgdi: Missing permission in java.policy.template added Review: AA HP-2008-02-07-0: Bugfix: fixed high CPU utilitzation by shepherd Cleanup: replaced shepherd_trace_sprintf and SHEPHERD_TRACE by an improved shepherd_trace function, replaced shepherd_error_sprintf, shepherd_error_impl and SHEPHERD_ERROR by an improved shepherd_error function. Changed: shepherd, qsh Bugster: 6659191 Review: MD MD-2008-02-07-0: Bugfix: o configfile check in automatic installation is to strict o certtool did not set and copy certs for WINDOWS installations Bugster: 6659855 Review: HP Changed: installation, certtool OJ-2008-02-07-0: Bugfix: Fix broken check in of OJ-2008-02-01-0 Issue: 1751 Reviewed: HP, AA OJ-2008-02-06-0: Bugfix: use of the same pathes for input/output stream must be dealt with Bugster: 6310194 Issue: 1751 Review: MP,RD AH-2008-02-04-0: Bugfix: Fixed shell error during inst_execd on (Darwin x86) Review: MD AA-2008-02-04-1: Bugfix: project.xml file of jgdi fix for SSLHelper warnings Review: RH RD-2008-02-04-1: CVS: Created V62alpha1_TAG ------------------------------------------------- EB-2008-02-01-0: Bugfix: Fixed bug which caused deadlock in master when scheduler thread was shutdown via qconf -ks Review: AH RD-2008-02-04-0: Bugfix: - fixed broken build on irix - fixed broken build for classic targets AA-2008-02-04-0: Enhancem.: SSL suport changes for JMX server - enable java.policy and GridwareConfig default login - adaption of install scripts and sge_ca script - correction for jvm thread startup in CSP mode Review: RH LP-2008-02-01-1: Bugfix: Fix for broken shell test on empty REMOVE_RC Review: PJ LP-2008-02-01-0: Enhancem: Improved qmaster installation with BDB server Enhancem: Cluster reinstallation can remove old cluster first Enhancem: Installation detects RC / SMF services for selected cluster name. Bugfix: Cluster name validation was not working Bugfix: darwin_template started qmaster before bdb Review: MD OJ-2008-01-31-0: Bugfix: execd installation does not test absolute path for local spool dir; Set SGE_QMASTER_PORT in settings file if sge_qmaster is not found in /etc/services file Bugster: 6528315, 6589297 Issue: 2195, 2477 Review: JG, MD JG-2008-01-31-0: Bugfix: Fixed memory leaks and qhost -j core dump. Review: RD HP-2008-01-31-2: Bugfix: Fixed some bugs in new IJS in shut down procedures Changed: qsh, shepherd Review: MD HP-2008-01-31-1: Bugfix: Fixed broken build on Irix65 after checkin RD-2008-01-31-0 Changed: sge_resource_quota_qmaster.c JG-2008-01-31-1: Bugfix: Fixed a memory leak in gdi. Review: EB AA-2008-01-31-0: Bugfix fixed RMON_USE_CTX again Review: EB RD-2008-01-31-0: Bugfix fixed redebitation if RQS gets enabled by -mattr Bugster: 6657521 Issue: 2476 Review: AH HP-2008-01-31-0: Bugfix: Fixed broken build on Irix65 after checkin EB-2008-01-30-1 Changed: sge_lock.c EB-2008-01-30-1: Bugfix: Enabled FIFO locking HP-2008-01-30-0: Bugfix: Fixed broken build on Irix65 after checkin EB-2008-01-30-0 Changed: sge_lock_fifo.c JG-2008-01-30-1: Bugfix: Fixed qmaster core dump. Review: AH AA-2008-01-30-0: Enhancem.: - broken build due to #if instead of #ifdef Review: EB CR-2008-01-30-0: Bugfix: - fixed commlib issue with message cleanup - code cleanup in tcp/ssl framework Review: RD, HP EB-2008-01-30-0: Enhancem.: - implemented FIFO lock as a replacement for pthread locking (pthread lock can be enabled by removing SGE_USE_LOCK_FIFO in libs/lck/sge_lock.h) - removed monitor. Now global locking is done earlier (old behaviour can be enabled with the preprocessor define DO_LATE_LOCK in libs/lck/sge_lock.h) Review: JG, CR JG-2008-01-30-0: Bugfix: qselect without options didn't show any queues. Review: AH RD-2008-01-30-0: Bugfix: fixed qquota for negated host or user scope Bugster: 6655195 Issue: 2472 Review: JG AH-2008-01-30-0: Bugfix: Healed reserveration scheduling: Parts of AH-2007-10-10-0 got lost due to the move of daemons/schedd/scheduler.c to daemons/qmaster/sge_sched_thread.c somewhen in the process of the scheduler-as-a-thread workpackage. Review: RD MD-2008-01-29-0: Bugfix: o qhost -l h= does not work Bugster: 4680766 Issue: 245 Changed: clients Review: JG RD-2008-01-29-0: Bugfix: fixed broken scheduler messages introduced with by performance enhancement Review: AH RD-2008-01-28-2: Bugfix: - fixed qdel force - corrected qrdel value parser Review: JG RD-2008-01-28-1: Cleanup: removed obsolete classic spooling code Review: AA LP-2008-01-28-0: Enhancem.: SGE_CLUSTER_NAME is valided before installation proceeds any further Enhancem.: qmaster service depends on bdb if installed Bugfix.: Execd uses libscf only on -ke request Review: RD RD-2008-01-28-0: Cleanup: removed drmaa 0.95 and drmaa-05.jar Bugster: 6655773 Issue: 2443 Review: AH MP-2008-01-25-0: Bugfix: Merged the changes from MP-2008-01-17-0 in V61_BRANCH to maintrunk Bugster: 6595754 Review: JG RD-2008-01-25-0: Enhancem.: lowered execd CPU overhead to improve job performance - PTF run is only triggered when priorization tickets arrive - added execd param PDC_INTERVAL to reduce PDC runs Review: JG LP-2008-01-25-0: Enhancem.: Fixed some typos and function description Bugfix: Fixed duplicate message id from LP-2008-01-24-0 Review: PJ LP-2008-01-24-0: Enhancem.: Subprocesses start in new contracts if SMF is used Enhancem.: Various SMF support improvements Enhancem.: Startup scripts now support -nosmf option Bugfix: load sensor might block execd port Issue: 2520 Bugster: 6654696 Review: RD MP-2008-01-24-0: Bugfix: Merged the changes from MP-2008-01-23-0 in V61_BRANCH to maintrunk Bugster: 6594807 Issue: 2465 Review: AH, JG AA-2008-01-24-0: Bugfix: del calendar can lead to SEGV Review: JG OJ-2008-01-23-0: Bugfix: 'NONE' as value is not rejected for queue_conf(5) shell and qsub(1) -S Bugster: 6453563 Issue: 2445 Review: RD, JG AA-2008-01-23-0: Enhancem.: - renaming AbstractEventClient->EventClientImpl, corr. cleanups - added nativeStartScheduler() - JMXEventMonitor smaller corrections - jgdish, qconf -sconf fix Review: RH JG-2008-01-23-1: Bugfix: Reintegrated fix for IZ 2050 (RN-2007-05-28-0). Review: AA JG-2008-01-23-0: Bugfix: fixed spooledit dump . Review: RD CR-2008-01-23-0: Bugfix: fixed commlib test binaries: test_cl_commlib test_cl_commlib_client Review: RD RD-2008-01-22-1: Bugfix - fixed load report issue due to improved qmaster-execd-protocol - fixed advance reservation soft error handling Bugster: 6652219, 6649605 Issue: 2469, 2470 Review: CR MD-2008-01-22-1: Bugfix: changed output string, which brakes testuite test Changed: install RD-2008-01-22-0: Bugfix: added cluster_name to backup files MD-2008-01-22-0: Bugfix: moved clear command which confused testsuite Enhancem.: added IsValidClusterName function to installer Review: JG Changed: install LP-2008-01-21-0: Bugfix.: Could not start install without SGE_ROOT set Bugfix.: Added missing SMF_FLAGS in the installation Review: PJ PJ-2008-01-21-0: Enhancem.: using INFOTEXT instead of echo Review: LP AA-2008-01-21-0: Doc: - man page additions in sge_conf(5) bootstrap(5) for scheduler thread and jvm thread Review: EB RD-2008-01-21-0: Bugfix: - corrected message queue cleanup - fixed event message handling for virtual clients Review: CR MD-2008-01-18-0: Bugfix: o autoinstall configfile should be parsed and checked for valid input! o automatic backup is broken Bugster: 6513435, 6550644 Issue: 2459, 2466 Changed: install Review: HP AA-2008-01-17-0: Enhancem: - event client hang fixes of jgdi client - NotificationBridge rewrite - AbstractEventClient rewrite adaption in jgdi_event.c and evc/sge_event_client.c - improvements in JGDIProxy and jmxeventmon - close method in JGDIJMXBean - exec host mod event is send if no last unheard - darwin x86 arch support in jgdi helper scripts Review: RH LP-2008-01-17-0: Bugfix: Fixed and improved LP-2008-01-16-0 Review: JG LP-2008-01-16-0: Bugfix: SMF was used in autoinstall with missing SGE_ENABLE_SMF Review: CR EB-2008-01-16-0: Enhancem.: Added missing qconf manpage entries Review: JG AH-2008-01-16-2: Bugfix: Parallel scheduling memory leak in master Bugster: 6562190 Issue: 2187 Review: JG HP-2008-01-16-1: Bugfix: Fixed broken build on irix65 after checkin HP-2008-01-16-0 Changed: sge_ijs_comm.c, sge_shepherd_ijs.c HP-2008-01-16-0: Bugfix: Fixed several bugs in new IJS regarding faulty pthreads implementation on some architectures, added commlib tracing to shepherd, improved error detection in communication. Changed: qrsh, shepherd Review: MD MP-2008-01-16-0: Bugfix: Merged changes of MP-2008-01-14-0 from the V61_BRANCH. Review: JG MD-2008-01-15-0: Bugfix: merged from V61_BRANCH MD-2007-10-04-0 some changes have been made due to differences between V61_BRANCH and maintrunk Review: JG Changed: commlib, gdi MD-2008-01-14-0: Bugfix: fixed commlib CRM message xml parser Review: CR Bugfix: fixed wrong inst_sge exit code Changed: commlib, inst_sge AH-2008-01-11-0: Bugfix: Inefficient single-job event commit pattern with qdel jobid1 jobid2 jobid3 Bugster: 6649547 Issue: 2458 Review: JG MD-2008-01-11-0: Bugfix: fixed memory allocation bug in getloadavg which was introduced with one of the last checkins Review: JG RD-2008-01-11-0: Bugfix: fixed protocol issue which lead a execd to consume 100% CPU time. Issue: 2457 Bugster: 6649528 Review: JG LP-2008-01-10-0: Enhancem.: Added correct permissions for SMF files to disinst Enhancem.: Improved sge_smf_support.sh Enhancem.: Removed version string v62 from the service names Enhancem.: Improved autoinstall log reporting. Enhancem.: Changed sgemaster (script) shutdown logic Bugfix: Fixed broken autoinstall due to LP-2008-01-04-0 Review: EB EB-2008-01-09-0: Enhancem.: improved shutown speed (pthread_cancel is sent earlier) Review: LP LP-2008-01-04-0: Enhancem.: Initial SMF support Review: JG AA-2008-01-04-0: Enhancem.: - jgdi_event.c functions moved to sge_event_client.c - cleanups in sge_event_client.c - sol-x86 libjvm fix - qconf -*conf, -sconf and -aconf fail and don't open editor - debugging messages if compiled with EVC_DEBUG - NotificationBridge.java subscribe before register fix - logging.properties.template cleanup Review: CR,EB JG-2008-01-03-0: Bugfix: accounting records for parallel job tasks didn't contain the task submission time. Bugster: 6469605, 6647681 Issue: 1689 Review: MD AP-2008-01-02-0: Bugfix: incomplete error loging in case of classic spooling failures Bugster: 6363756 Issue: 2442 Review: JG JG-2007-12-31-0: Bugfix: Fixed broken CL_SIRM_MESSAGE definition. Review: TBD JG-2007-12-28-0: Enhancem.: Added option to suppress individual accounting records for tightly integrated parallel jobs. Bugster: 6572326 Issue: 2451 Review: HP JG-2007-12-27-0: Bugfix: Fixed duplicate message ids. RD-2007-12-23-0: Bugfix: queue changes with an AR generated a lot of unnecessary events Review: TBD RD-2007-12-22-0: Enhancem.: - optimized execd data collector execution - qsub now allows comment after script directive Review: TBD HP-2007-12-21-2: Enhancem.: Installation changed so that the new interactive jobs are now used by default. Changed: inst_sge, arch_variables, inst_common.sh, inst_qmaster.sh Review: MD HP-2007-12-21-1: Bugfix: Fixed wrong library path in settings.[c]sh file on HP1164 Changed: dist/util/create_settings.sh Review: MD HP-2007-12-21-0: Enhancm.: o Activated csp mode in new interactive job support o Fixed broken build with aimk -no-secure Review: MD RD-2007-12-21-0: Enhancm: commlib scalability enhancments - added hashed access for connection list - added hashed access for endpoint list - added hashed access for host list - reduces protocol overhead by removing unused fields - optimized memory usage Review: CR CR-2007-12-20-1: Bugfix: o fixed some ant build targets in build.xml files Review: AA CR-2007-12-20-0: Bugfix: o fixed some ant build targets in build.xml files Review: AA AA-2007-12-17-0: Bugfix: - subscription before eventClient.start() fails - typos Review: RH JG-2007-12-17-0: Enhancem.: higher precision for cpu, ru_utime, ru_stime Bugster: 6288965 Issue: 869 Review: HP RH-2007-12-17-0: Enhancem.: o JGDI JMX authentication improved o docu for sample jaas.config added Review: AA MD-2007-12-14-0: Enhancem.: o switched gelint script from lint to splint o script enhanement RH-2007-12-14-0: Enhancem.: o JGDI jmx session created - JGDILoginModule creates the session id - JGDIAgent registers for each session a mbean - correct connection shutdown handling - only the owning connection can access this mbean o Bugfixes in NotificationBridge and JGDProxy o password less login with keystores o sge_ca creates keystores for users and copies user certificates into $SGE_ROOT/$SGE_CELL/common/sgeCA/usercerts Review: AA PJ-2007-12-13-0: Bugfix: ST, mk_dist should remove sge_st_settings.sh instead of distinst Bugfix: ST, sundist does not have needed st_settings.sh Review: RD EB-2007-12-13-0: Enhancem.: added possibilitie to start and stop scheduler and jvm thread Review: MD RD-2007-12-13-0: Bugfix: Fixed some more memory leaks introduced with the scheduler thread, bdb jobscript spooling and the abstract event client Review: CR HP-2007-12-12-0: Bugfix: Installation failed on Windows when installing as root without SGE admin user Changed: dist/util/install_modules/inst_execd.sh Bugster: 6640885 Review: RD JG-2007-12-12-0: Bugfix: JG-2007-12-10-0 introduced too strict verification of the JB_cwd job attribute. Review: RD AA-2007-12-11-0: Bugfix: - AbstractEventClient fix - evc/evm fixes - jgdi_qstat -l option fix - build.xml dependencies - jgdi qconf -mq, h_rss etc. verify Review: RH JG-2007-12-11-0: Bugfix: Replace realloc by sge_realloc, make sure that memory is freed, when realloc fails. Review: RD RD-2007-12-11-0: Bugfix: fixed some memory leaks introduced with the scheduler thread Review: EB JG-2007-12-10-0: Bugfix: Added missing verification to pathes in cluster configuration. Bugster: 6506580 Issue: 2136 Review: MD PJ-2007-12-07-0: Bugfix: The Service tags settings should differ for sundist/mk_dist Bugfix: The error when no ARCH was defined. Review: LP RD-2007-12-07-0: Bugfix: - fixed issue with unheard execd and diffed load reports - corrected weird compare for reprioritze at qmaster startup - fixed possible segfault with commlib params Review: CR AH-2007-12-06-0: Bugfix: Use of function pointers in dtrace pid provider functions broke master monitor Bugster: 6638313 Review: RD HP-2007-12-05-0: Bugfix: fixed broken build with -no-jni option Changed: sge_c_gdi.c Review: AA AA-2007-12-05-0: Bugfix: fixed accidental checkin of sge_log.c breaking qrsh output Review: RD RD-2007-12-05-0: Bugfix: fixed qping core dump Review: MD AH-2007-12-04-0: Enhancem: Merged -hold_jid_ad related man page changes from V61u1_ARI_BRANCH Bugster: 6367649 Issue: 1832 Review: RD EB-2007-12-03-1: Bugfix: o order number in profiling output of scheduler was not correct Review: JG RD-2007-12-03-0: Bugfix: fixed broken build on darwin Review: EB EB-2007-12-03-0: Bugfix: o fixed bug in scheduler which might cause core of scheduler if it is restarted. o fixed bug which might cause core in logging code when scheduler was shut down. o fixed bug which caused wrong runtime values in throuput tests o reworked startup/shutdown code for scheduler and jvm thread in qmaster (client side is still missing) Review: AA MD-2007-10-04-0: Enhancem.: commlib enhancement, fix is still missing for this CR clients should issue a more explicit error message when qmaster is busy Bugster: 6469494 Issue: 2372 Changed: commlib, spooling RD-2007-12-03-0: Enhancem.: improved execd startup Review: CR AA-2007-11-30-0: Enhancem.: - broken jgdi qstat output fix - internal event client for jgdi - update_func moved to local context - support of solaris jvms on solaris8/9 link against -lthread instead of lpthread which prevented start of jvm_thread Review: RD JG-2007-11-30-0: Enhancem.: Set default for reporting_params log_consumables to resource friendly default "false". Some cleanup. Review: RD MD-2007-11-29-0: Bugfix: o the installation of two rpc databases on the some host fails Bugster: 6259265 Issue: 2437 Changed: install Review: JG AA-2007-11-29-0: Bugfix: inst_qmaster.sh adding of -d64 on sol-sparc64 and sol-amd64 prevents start of jvm_thread Review: RD JG-2007-11-29-0: Bugfix: Verify slots attribute in cluster queue and pe. Bugster: 4752829 Issue: 2433 Review: RD RD-2007-11-29-0: Bugfix: fix 100% CPU usage qmaster if JVM thrad couldn't started Review: AA LP-2007-11-27-0: Bugfix Fixed some JGDI Shell options Bugfix Marked lt_heard_from as hidden in JGDI Review: AA RD-2007-11-26-1: Bugfix: disabled starting of sge_schedd by sge_shadowd Review: EB RD-2007-11-26-0: Bugfix: Fixed long qmaster shutdown time Review: CR JG-2007-11-26-0: Bugfix: Fixed minor memory leaks. Review: EB EB-2007-11-23-1: Bugfix: Fixed broken scheduler due to EB-2007-11-23-0 AH-2007-11-23-0: Bugfix: Fix broken qstat -s a caused by AH-2007-11-16-0 Copyright: Fixed 'Initial Developer' statement in sge_task_depend.[ch] Review: RD EB-2007-11-23-0: Cleanup: o Moved sge_share_mon from clients to utilbin o Added parameters to bootstrap file to enable threads or set the initial number of threads o Added qmaster param temporaryly to limit the number of GDI order requests send by qmaster o Changed the initial number of worker an listener threads o Eleminated uneccessary pthread_cond_wait() call in worker threads which reduced overall performance. Review: JG NOTE: Reinstallation necessary or otherwise scheduler and JVM thread won't start during qmaster start. or Add following lines to your bootstrap file: listener_threads 2 worker_threads 2 scheduler_threads 1 HP-2007-11-23-0: Bugfix: new IJS: fixed broken environment setup for qrsh without command when environment of shepherd is incomplete. RH-2007-11-22-0: Bugfix: broken jgdi build fixed Review: AA AA-2007-11-22-0: Enhancem: o better logging for JGDIAgent o own MBeanServer instead of Platform MBeanServer Review: RH RD-2007-11-22-0: Enhancem: added qconf -Mconf Bugster: 6609633 Issue: 2422 Bugfix: - fixed wrong exit code for qconf -sprj - fixed schedd core dump because of hostname resolving Review: AA, AH CR-2007-11-22-0: Bugfix: not initalized variable caused qmaster segfault on solaris amd64 hosts Review: AA PJ-2007-11-21-1: Bugfix: ExecuteAsAdmin shorten the long command HP-2007-11-21-0: Enhancem.: o new interactive job support o starts interactive jobs directly, without rsh/rshd or rlogin/rlogind or telnet/telnetd. o doesn't need reserved ports on exec host. o redirects input/output using commlib, i.e. this redirection is prepared for encryption. Added: clients/common/sge_client_ijs.c clients/common/sge_client_ijs.h common/sge_ijs_comm.c common/sge_ijs_comm.h common/sge_ijs_threads.c common/sge_ijs_threads.h daemons/shepherd/sge_shepherd_ijs.c daemons/shepherd/sge_shepherd_ijs.h libs/uti/sge_pty.c source/libs/uti/sge_pty.h Modified: aimk common/Makefile clients/qsh/Makefile clients/qsh/msg_qsh.h clients/qsh/qsh.c common/Makefile daemons/qmaster/sge_thread_jvm.c daemons/shepherd/Makefile daemons/shepherd/builtin_starter.c daemons/shepherd/shepherd.c libs/uti/Makefile utilbin/Makefile Review: JG PJ-2007-11-21-0: Bugfix: ExecuteAsAdmin shorten the long command line arguments an solaris host because of /usr/bin/sh limitations Review: AS AH-2007-11-21-0: Bugfix: Completized fix of dtrace monitor after EB-2007-11-02-1 reconstructions Review: RD AA-2007-11-21-0: Enhancem.: o jvm logging/startup o qmon qconf jvm option support o is_qmaster_internal_client in ctx to define if sge_evc is internal instead of extra local_client flag o SKIP_NONREENTRANT in aimk and scripts/nonreentrant.sh for insure Review: RH RH-2007-11-21-0: Bugfix: o JGDIAgent calls now AbstractEventClient.closeAll and JGDIBaseImpl.closeAll at shutdown o UnixLoginModule does not call authuser if no username is specified Review: AA LP-2007-11-20-0: Enhancem.: Removed Command, HistoryCommand interfaces from JGDIShell Enhancem.: JGDIShell now stores original argument lists Enhancem.: JGDI project attributes visibility changed from RO to HIDDEN Bugfix : Fixed some command options and messages in JGDI Review: AA DT-2007-11-19-0: Bugfix: o Corrected DRMAA Java version number o Added -sync to list of native switches removed by DRMAA o Added libdrmaa95.so to drmaa_clean target Review: pending Bugster: 2253, 2430 Issue: 6558329, 6631777 RD-2007-11-19-0: Enhancem: added README-Upgrade.txt to distribution Review: JG AA-2007-11-19-0: Enhancem: export SGE_ENABLE_JMX for bootstrap file Review: EB AH-2007-11-16-0: Enhancem: Merged -hold_jid_ad option from V61u1_ARI_BRANCH Copyright: Added Rising Sun Pictures copyright Review: CR EB-2007-11-15-1: Bugfix: Fixed bug in scheduler (queue selection) Removed incorrect logging during scheduler initialisation AA-2007-11-15-0: Enhancem.: problem on sol-amd64, sol-sparc64 jvm_thread -d64 option added in install Review: CR CR-2007-11-15-0: Bugfix: added chmod for sge_st script in distinst Review: AA EB-2007-11-15-0: Bugfix: Added error logging. removed unused MSG_ defines EB-2007-11-14-0: Bugfix: Suspend thresholds were broken. Review: JG (AH) Bugfix: host_list_locate() used with empty execd, admin, or submit hostlist caused error logging in message file Review: JG Bugfix: jobs where scheduled twice or deleted twice due to error in busy state handling of schedd Review: JG CR-2007-11-14-0: Bugfix: authuser binary checked real user id not the effective user id. Problem: Only root could start this binary. Review: RH AA-2007-11-14-0: Enhancem: - configuration of jvm thread bootstrap->global/local configuration - shutdown of jvm thread - fix for qmon new data fetching - fix for 100% CPU usage in empty jvm thread Review: RH AH-2007-11-12-0: Enhancem: Need GNU-lib mtrace()/muntrace() support in qmaster for hunt on memory leaks Bugster: 6627903 Issue: 2426 Review: HP MP-2007-11-09-0: Bugfix: Fixed the break introcuded in testsuite test from the commit MP-2007-10-26-0 Review: JG OJ-2007-11-06-1: Bugfix: Fixed minor issues in man pages Review: OJ-2007-11-08-0: Bugfix: Fix qconf -mc accepts erroneous resource entries without an urgency; qmon gives (poor) error message Bugster: 6553062 Issue: 2383 Review: AH AH-2007-11-08-0: Bugfix: Part of AH-2007-10-31-0 was lost due to EB-2007-11-02-1 Review: MD AH-2007-11-07-1: Bugfix: EB-2007-11-02-1 broke dtrace monitoring Review: EB AH-2007-11-07-0: Bugfix: EB-2007-11-02-1 broke the schedule file Review: EB EB-2007-11-07-0: Bugfix: o worked on open TODO's due to review process of for work package EB-2007-11-02-1 (only cleanup) Review: MD RH-2007-11-05-0: Bugfix: o broken build on aix43 o jvm thread is built if NO_JNI is not defined Review: AA AA-2007-11-05-0: Bugfix: o fix for broken jgdi_qstat due to AH-2007-10-31-0 EB-2007-11-05-1: Bugfix: o merged missing code concerning JVM instatiation. Review: RH EB-2007-11-05-0: Bugfix: o Renamed preprocessor variable to enable the JVM thread code to "ENABLE_JVM_THREAD" EB-2007-11-03-0: BugFix: o Fixed bug in event master introduced with EB-2007-11-02-1. EB-2007-11-02-2: BugFix: o Disabled qstat enhancement checked in with AH-2007-10-31-0. (qstat -f did not work) o Fixed source/dist/util/sgeST/sge_st which broke installation. EB-2007-11-02-1: Enhancem.: o scheduler as a thread in master o possibility to use GDI within master threads o hooks for the event/mirror interface to bypass communication and packing in master threads o replaced all threads in master by "commlib-threads" o reworked steps done during startup/shutdown of master Review: JG, AA EB-2007-11-02-0: CVS Created EB_PRE_SAAT_TAG --------------------------------------------------- AH-2007-10-31-0: Bugfix: - reduce amount of queue list copying in scheduler - care for propagation of queue suspension state to jobs in qmaster as to minimize data amount to be fetched by clients - add support for queue suspension to SIMULATE_EXECD mode Bugster: 6595856 Issue: 2350 Review: MD JG-2007-10-31-0: Bugfix: Merge from V61_BRANCH of JG-2007-10-30-1: qstat -ext didn't show the usage of parallel tasks in the summary view (without -g t) Bugster: 6391244 Issue: 2419 Review: RD Merge from V61_BRANCH of JG-2007-10-31-0: Bugfix: ARCo ju_exit_status should provide means to recognize the intermediate record. Bugster: 6621482 Issue: 2416 Review: RD PJ-2007-10-31-0: BugFix. : Needs to be executable file PJ-2007-10-30-0: Enhancem.: The GE Service Tags support Review: RD JG-2007-10-30-0: Bugfix: qrsh -no now set all queues in error state when the qrsh was unavailable when the job was scheduled. Bugster: 6588743 Issue: 2418 Review: HP, EB AA-2007-10-29-0: Enhancem: o jvm_thread fix for using certs for JGDIAgent o multiple mbean instances o JGDIShell problem for macos o tpyo fix jass -> jaas, update your ClusterConfig*.properties *jass*->*jaas* Review: RH JG-2007-10-29-0: Bugfix: The start_time field was incorrect in intermediate accounting records. Bugster: 6622842 Issue: 2417 Review: RD MD-2007-10-26-0: Bugfix: o Incorrect messages in qconf command, o rctemplates lack of requirement, o the restore feature does not delete old configuration before restoring Bugster: 6596931, 6287501, 6618619 Issue: 2406, 2336, 2407 Changed: clients, install, backup_restore Review: HP MP-2007-10-26-1: Bugfix: sge_shepherd should exit if it cannot write to any of its essential files Bugster: 6564543 Issue: 2400 Review: JG MP-2007-10-26-0: Bugfix: qstat -j -xml shows more detailed information than qstat -j Bugster: 6353628 Issue: 1909 Review: JG RH-2007-10-26-0: Enhancem: Installer enhancement for JMX MBean Server added Review: MD RD-2007-10-25-0: Bugfix: - fixed execd reconnect on qmaster migration - fixed incorrect error reporting for job script spooling Review: JG JG-2007-10-24-1: Bugfix: Long running jobs caused incorrect usage information from ARCo database. Bugster: 6618599 Issue: 2414 Review: RD JG-2007-10-24-0: Bugfix: Fixed memory leaks / memory access errors. Review: RD RD-2007-10-24-0: Bugfix: raised GDI version JG-2007-10-22-0: Bugfix: add a parameter to reporting_params to switch off writing of all consumables Bugster: 6617450 Issue: 2401 removing parameters from the reporting_params will not fallback to default values Bugster: 6619016 Issue: 2409 Review: EB HP-2007-10-22-0: Bugfix: Fixed broken build after checkin AH-2007-10-22-1 Changed: clients/common/sge_qhost.c AH-2007-10-22-1: Bugfix: Further reduction with orphan queue checking overhead and qhost optimization. Bugster: 6619657,6619662 Issue: 2410,2411 Review: EB RD-2007-10-21-0: CVS Created V61ARsnapshot3_BRANCH --------------------------------------------------------- RD-2007-10-21-0: CVS Created V61ARsnapshot3_TAG ------------------------------------------------------ AH-2007-10-19-1: Bugfix: Bypass expensive sge_print_queues() if not needed in qhost Bugster: 6619662 Issue: 2411 Review: RD AH-2007-10-19-0: Bugfix: Bypass expensive but unneeded orphan queue checking in qmaster upon qmod -d|-e|-s|-us Bugster: 6619657 Issue: 2410 Review: RD LP-2007-10-19-1: Enhancem.: JGDIShell now uses stderr to display errors Enhancem.: JGDIException now has exitCode Bugfix.: JGDI qconf -ae now has the same behaviour as client Review: AA AH-2007-10-18-0: Bugfix: Queue creation time grows more than linear with large clusters Prevent "missing slots attribute" logging upon queue instance creation Bugster: 6618565 Issue: 2405 Review: RD AA-2007-10-18-0: Bugfix: qmon segfault functional/override policy qmon queue customize corrupted list string jgdi shutdown global variable link dependency long/short hostnames in build.xml Bugtraq: 6618315, 6618323, 6618328 Review: RD RD-2007-10-18-0: Bugfix: timed event thread didn't set unknown host into unknown state Review: JG LP-2007-10-17-1: Enhancem.: JGDIShell returns correct exitCodes in most cases Review: AA HP-2007-10-17-0: Bugfix: Fixed minor bugs, compiler warnings and comments Changed. cl_commlib.c, sge_prog.h, cl_log_list.c, str2nm_converter.c, execd_tickets.c, execd_job_exec.c, msg_gdilib.h, sge_qexec.c Review: MD RD-2007-10-17-0: Bugfix: load sensor complex did not disappeared introduced by RD-2007-10-08-0 Review: JG AH-2007-10-16-0: Bugfix: Fixed issue with round_robin host consumable debitation caused by AH-2007-10-10-0. Review: RD RD-2007-10-16-0: Bugfix: - fixed bugs of new parallel reservation code - reverted part of qmaster performance improvements from RD-2007-10-12-0 JG-2007-10-15-0: Bugfix: Scheduling of parallel jobs did not respect consumables, if the consumables were referenced in a resource quota Bugster: 6608236 Issue: 2396 Review: RD RD-2007-10-12-1: Bugfix: fixed AR scheduling bug due to AH-2007-10-10-0 Review: AA RD-2007-10-12-0: Enhancem. minor qmaster performance improvements Review: AA AH-2007-10-10-0: Cleanup: - integrated parallel two-stage slot selection into a single stage as to reduce dispatch time Bugfix: - -masterq may not break slave task preference due to soft/queue sort method - resource quota can break allocation_rule - fixed issue reservation until infinity in parallel scheduling Enhancem.: - need means to simulate large numbers of execution nodes Bugster: 6610855,6586817,6615165 Issue: 2378,2339,2364 Review: RD JG-2007-10-09-1: Bugfix: o sharetree.5 didn't explain type field. Bugster: 6575731 Issue: 2276 o scheduler printed empty lines in messages file Bugster: 6608259 Issue: 2394 AA-2007-10-09-0: Bugfix: Fixed qmon binary job submit is broken jgdi cleanups: host name issues for -sconf, -se sharetree handling dump method corrections in filters qstat load_avg qhost filters Bugster: 6604155 Issue: 2346 Review: RD JG-2007-10-09-0: Bugfix: Fixed gcc-4.2 warnings, minor bugs, cleanup. Review: MD RD-2007-10-08-0: Enhancem: improved qmaster-execd-protocoll - enabled threaded commlib for execd - rewrite of execd dispatcher - only changed load reports are now send to qmaster Cleanup: - cleanup qdel and the job del code in qmaster Issue: 664 Review: AA RH-2007-10-05-0: Bugfix: - high cpu utilization in jvm thread fixed - missing functions in JMX JGDI added Review: AA LP-2007-10-04-2: Cleanup: Reformatted JGDI java files to 4 spaces Review: LP LP-2007-10-04-1: Following reffer to JGDI and JGDShell only Enhancem.: Added new error messages to resource file Bugfix.: Fixed some error messages for qconf Review: AA LP-2007-10-04-0: Following changes reffer only to JGDI and JGDShell Enhancem.: qstat -g c [ext] recognized Enhancem.: Added resource file with error messages Bugfix.: Fixed many error messages for qconf and qstat Review: AA RN-2007-10-03-0: Bugfix: SGE does not work if primary group entry is too big in groups map Bugster: 6568575 Issue: 2249 Review: AH HP-2007-10-01-0: Bugfix: Fixed broken build after checkin AH-2007-09-28-0: Removed dependency to japi.1.0.o from qsub Makefile Changed: clients/qsub/Makefile AH-2007-09-28-0: Bugfix: Fix unneeded builds of JAPI targets due to AH-2007-09-27-0 by using japi.o instead of japi.1.0.o, drmaa.o instead of drmaa.1.0.o etc. RD-2007-09-28-0: Bugfix: - fixed qrstat -ar for AR of different user - enhanced job scheduling for AR jobs to allow the request of non consumable resources - implemented qrstat default request file Bugfix: - wc_ar_list is not explained in sge_types man page - fixed qrdel as admin for ARs of different user - fixed qdel exit code handling Bugster: 6597428, 6597549, 6610788 Issue: 2376, 2377, 2375 Review: AA JG-2007-09-28-0: Bugfix: Installation failed on Linux with glibc 2.6 Bugster: 6594665 Issue: 2352 Review: RD JG-2007-09-27-0: Bugfix: Fixed man page issues. Bugster: 6575727, 6575720, 6589459 Issue: 2272, 2270, 2374 Review: RD RD-2007-09-27-0: Bugfix: - fixed broken build on hp11 - fixed broken sge_depend with SS12 Review: AA AH-2007-09-27-0: Bugfix: Fix broken dependencies of japi.1.0.o and japi.0.95.o build targets Review: JG OJ-2007-09-26-0: Bugfix: Unset old_value out of scope Bugster: 6569862 Review: MD RN-2007-09-26-0: Bugfix: problem with floating point job resource limits Bugster: 4743006 Review: JG MP-2007-09-25-0: Bugfix: qstat -j messages disagree between plain and XML output Bugster: 6440408,6440412 Issue: 2076 Review: RD AA-2007-09-25-0: Cleanup: - typed lists and other 1.5 cleanups Review: LP AH-2007-09-24-1: Bugfix: Fix schedd crash with test case 2300 due to AH-2007-06-15-0 AH-2007-09-21-0: Bugfix: Resource reservation broken with sequences of identical jobs differing only in their -R y|n Bugster: 6590079 Issue: 2344 Review: RD MP-2007-09-21-0: Bugfix: qdel does not recognize wc_job_range_list as it is defined. Bugster: 6597547 Issue: 2367 Review: JG RN-2007-09-21-0: Bugfix: - authuser binary returns unusable error message - Several qconf options display only single message when a list of messages should be printed Bugster: 6527836,6577034 Issue: 2194,2369 Review: JG RD-2007-09-21-0: Bugfix: fixed broken build on hp11 Review: JG AH-2007-09-19-1: Bugfix: Fix scheduling of PE range jobs when resource quotas are used. It was broken by OJ-2007-09-10-0 which in turn was broken by AH-2007-06-15-0 when resource quota result caching was introduced for parallel jobs. Review: RD RH-2007-09-19-1: Enhancem.: installation copies jmxremote.password file to $SGE_ROOT/$SGE_CELL/common and sets the correct permissions Review: AA RH-2007-09-19-0: Enhancem.: - JGDI: Dynamic proxy for accessing JGDI over JMX - qmaster: jvm thread is enabled if env variable SGE_LIBJVM_PATH is set Review: AA RN-2007-09-12-0: Bugfix: - Creating a userset with NONE as a type results in a core dump - Userset spooling in classic mode is broken Bugster: 6576153,6600619 Issue: 2323,2356 Review: RD LP-2007-09-07-0: Enhancem.: JGDIShell detects multiple option registrations JGDIShell can handle exit codes Cleanup: JGDIShell commands Review: PJ AA-2007-09-06-1: Bugfix.: - JGDIRemoteCacheListener readded AA-2007-09-06-0: Enhancem.: - commlib read error messages fix in sge_log() - missing answer lists in QConfCommand - sge_edit() external editor handling - deletesWithAnswer() method - QDelCommand, QModCommand - replacement of java_synchronized_jgdi.jsp via proxy class - StringBuffer -> StringBuilder Review: RH PJ-2007-09-06-0: Enhancem.: added qsub command to JGDI Cleanup: qrstat, qrsub Review: LP RH-2007-09-05-0: Enhancem.: jgdi event support for jmx interface added Review: AA RN-2007-09-04-0: Bugfix: qconf -(A,D,M,R)attr dumps core when the supplied file is empty Bugster: 6578213 Issue: 2327 Review: JG LP-2007-09-03-0: Enhancem.: Added new Util class Bugfix: Fixed several broken JGDIShell command options Review: PJ PJ-2007-09-03-0: Enhancem.: Implemented qrstat, qrsub in JGDI Cleanup: Command refactoring Review: LP CR-2007-08-31-0: Cleanup: sge_ca script now does umask 077 also for no root user installations Review: AA MP-2007-08-31-0: Bugfix: qstat -j output disagrees between xml and plain messages Bugster: 6440408 Review: JG AA-2007-08-30-0: Enhancem.: - bugfix for AA-2007-08-20-0 - N1GE->SGE - test_ff_cl - message numbers Review: RD RD-2007-08-30-0: Docs: added ARCo Space Requirements HowTo Review: JO LP-2007-08-29-2: Enhancem.: JGDIShell uses single PrintWriter All JGDIShell -help options use UsageResources file Review: PJ RD-2007-08-29-1: Bugfix: fixed broken build on win32-x86 Review: AH AH-2007-08-29-1: Bugfix: qsub -t 1-N:N creates a normal job with one task Bugster: 6597463 Issue: 2351 Review: EB LP-2007-08-29-1: Enhancem.: Using LONG32_MAX Added missing copyrights Review: PJ RD-2007-08-29-0: Bugfix: fixed issue with SS12 and dtrace on amd64 Review: AH LP-2007-08-29-0: Enhancem.: Added JGDI QrStat, Qrdel cmds Enhancem.: Now using file imports in JGDI generation process (*.static files) Bugfix: Fixed JGDI qconf -acal Bugfix: Handling ulong32 to jint overflow in JGDI Review: PJ RH-2007-08-27-0: Bugfix: juti: Commit method of UnixLoginModule does not report RuntimeExceptions Bugster: 6597423 Issue: 2353 Review: JG LP-2007-08-24-1: Bugfix: Quick fix for JGDI build break Review: LP LP-2007-08-24-0: Cleanup: Merged new JGDI *WithAnswer functionality Code cleanup using Java 1.5 Enhancem.: Using annotation and reflection to cache option methods in QConfCommand. Introduced UsageResource bundle for the commands Generating generic qconf option methods using new java_qconf_cmd.jsp template Review: PJ EB-2007-08-21-0: Build: Fixed broken installation when classic spooling was used Review: AA AH-2007-08-20-1: Build: Build fails with bleeding edge glibc Issue: 2348 AA-2007-08-20-0: Enhancem.: - require jdk 1.5 for jgdi - comment cleanup - userprj UP_Type split into UU_Type and PR_Type - manager/operator MO_Type split into UM_Type and UO_Type - added *WithAnswer methods to return answers from master in jgdi - jvm thread code: to activate -> aimk -DJVM_THREAD in bootstrap file add: libjvm_path /jre/lib//server/libjvm.so Review: EB MD-2007-08-13-0: Bugfix: merged checkin MD-2007-07-23-0 from V61_BRANCH LP-2007-08-09-0: Enhancem.: QConfCommand restructured. Enhancem.: Java templates have now specific code for some classes Review: AA DT-2007-08-09-0: WWW: o Fixed broken download link AH-2007-08-09-0: Bugfix: Resource quota can prevent dispatching of jobs that requests no resource in this quota Bugster: 6590574 Issue: 2345 Review: RD MD-2007-08-08-0: Enhancem.: - changed gelint and aimk due to daily build integration HP-2007-08-07-0: Bugfix: - Overflow of group members in getgrgid_r() call on Interix causes submit clients and execd to fail. - Fixed broken build. Bugster: 6589600 Changed: libs/uti/sge_uidgid.c Review: JG JG-2007-08-07-0: Bugfix: newline missing from "illegal debug level" message Bugster: 6589807 Issue: 2343 Review: MD AH-2007-08-03-0: Bugfix: Fixed issue with parallel RQS result caching introduced by AH-2007-06-15-0 with multiple and craggy quota set-ups Review: RD MD-2007-08-02-0: Enhancem.: added gelint script. (scripts for source code checking) Review: JG Changed: aimk, /scripts/scripts.html PJ-2007-08-01-1: Bugfix: DTRACE fix Changed: The monitor.d script and signatures of job_write_script, job_read_script, job_delete_script Review: LP PJ-2007-08-01-0: Enhancem.: Bdb jobscript spooling Review: RD HP-2007-07-31-0: Bugfix: Fixed broken build after checkin RD-2007-07-26-0 Changed: source/libs/spool/Makefile AH-2007-07-27-0: Bugfix: Parallel RR broken if jobs wait for queue slots and no RQS configured Bugster: 6585721 Issue: 2338 Review: RD RD-2007-07-27-0: Bugfix: removed duplicate dlopen() call in libspoold Review: AA RD-2007-07-26-0: Cleanup: - Made the flatfile spooling framework as the default for classic spooling - unified classic spooling for advance reservation - fixed broken pe range spooling for advance reservation Review: AA HP-2007-07-24-0: Bugfix: Fixed broken build on irix65 after checkin GH-2007-07-24-0 Review: AH AH-2007-07-24-0: Bugfix: user/system/operator hold state combinations cause strange qstat output Bugster: 6584632 Issue: 2334 Review: JG GH-2007-07-24-0: Bugfix: qmaster failover should not change the state of any queue Bugtraq: 6396036 Review: AH,JG RD-2007-07-24-0: Bugfix: fixed compile error on AIX51 Review: AA MP-2007-07-19-0: Enhancem.: Checked spelling of all man pages. Review: OJ RD-2007-07-18-0: Bugfix: - fixed AR-job scheduling with resource reservation enabled - enabled -R y for AR-jobs - fixed issue with multiple -l resource requests with qrsub Review: AH PJ-2007-07-18-0: Bugfix: fixed (non critical) -ar switch for qsh, qrsh, qlogin Review: JG DT-2007-07-16-0: Tests: o Fixed bad test formulation in DRMAA junit test Review: JG Changed: DRMAA junit tests JG-2007-07-16-0: Bugfix: fixed (non critical) memory leaks Review: RD AH-2007-07-11-0: Bugfix: Various improvements with sequential assigments and reservation: - kicked out per queue instance soft violation caching - cache operation results and static host matching right after CQ matching - added per host -q evaluation - stop iterating through queue instances when a soft/ reservation solution is determined that won't become better - once a soft/reservation solution is found, queue instances no longer are investigated further, as soon as it becomes apparent no better soft/reservation solution will jump out - do RQS shadowing also for reservation jobs if they hinder better solution than already found - do host/cqueue exclusion with RQS shadowing also when host and cqueue shadows global limit rule Bugster: 6579232 Issue: 2328 Review: AA RN-2007-07-11-0: Bugfix : - Creating a userset with NONE as a type results in a core dump - Userset type should not accept empty string value Bugster: 6576153, 6576197 Issue : 2323, 2324 Review : JG RD-2007-07-09-0: Bugfix: fixed broken RQS-AR interaction Review: JG MD-2007-07-06-0: Bugfix: sge_conf(5), description for rlogin_daemon and qlogin_daemon is wrong Bugster: 6506661 Issue: 2326 Review: JG JG-2007-07-06-0: Bugfix: Fixed memory leaks in qsh. Review: RD JG-2007-07-05-0: Bugfix: Fixed get_pw_buffer_size and get_group_buffer_size. Fixed drmaa error when native specification contained leading spaces. Bugster: 6542987 Issue: 2325 Review: RD AP-2007-07-04-0: Bugfix: qacct -l = ignores jobs in output when they reference no longer existing queue. Instead of warning message for each job added one message with a count of ignored jobs. Bugster: 6525375 Issue: 2189 Changed: qacct, sgq_cqueue, msg_history Review: JG RD-2007-07-04-1: Bugfix: Fixed broken job runtime introduced by RD-2007-06-29-0 Review: AA RD-2007-07-04-0: Bugfix: Fixed broken resource reservation after commit RD-2007-06-29-0 Review: JG JG-2007-07-03-0: Bugfix: Fixed memory leaks in event master. Fixed memory leaks in qconf. Review: RD AA-2007-07-02-0: Bugfix: - in qmon, under calendar configuration, it is possible to modify even if no calendar exists - Inadequate error message when qconf -sstree is run when no share tree is configuredIf no share tree fixes for answer_list handling - jgdi fixes for Configuration, Manager, Operator Bugster: 6513115, 6571714 Issue: 2320, 2321 Review: RD DT-2007-07-02-0: Bugfix: o Last fix for license text for JavaDocs Changed: build.properties DT-2007-06-29-0: Bugfix: o Fixed license text for JavaDocs Changed: build.properties AA-2007-06-29-0: Bugfix: incomplete fix of AA-2007-06-18-0 /etc/services worked, set ports didn't Review: MD RD-2007-06-29-0: Bugfix: - qstat didn't showed the current reserved slots - fixed acl handling for advance reservation - duration offset was not enforced for AR jobs - duration offset config was ignored - fixed masterq scheduling for AR jobs Review: PJ DT-2007-06-28-0: Enhancem.: o Added license text to JavaDocs o Added missing Changelog history Changed: build.properties, classes/build.xml AH-2007-06-28-0: Bugfix: - parallel resource reservation broken when non-queue instance based quotas limits apply - sort out queues/hosts unsuited due to RQS - call job_has_soft_requests() only once per assignment - move RQS scheduling code from sge_select_queue.c into sge_resource_quota_schedd.c Bugster: 6571749 Issue: 2303 Review: AA RN-2007-06-28-0: Bugfix: qconf -sq all.q@myhost produces no value at all for complex_values (not even NONE) Bugster: 6433628 Issue: 2307 Review: JG RD-2007-06-27-0: Bugfix: added -m64 in aimk for sol-amd64 and SunStudio 12 Review: MD JG-2007-06-26-1: Bugfix: qrsh with command reported an error when ssh was configured as transport, and CTRL-C was pressed. Bugster: 6573508 Issue: 2316 Review: RD JG-2007-06-26-0: Bugfix: Fixed (non critical) memory leaks. Review: RD AA-2007-06-26-0: Bugix: - Qmon Job Control Window: Incomprehensible Priority Button - Double clicking in Consumables/Fixed Attributes list... - complex attributes whose deletion is denied... - Help for Browser panel in qmon incomplete - Qmon x qconf inconsistent in allowed characters in attribute names - Qmon panel does not check for valid data in Scheduler Configuration - It is possible to negative tickets / shares in qmon and from the command line (partially fixed CR 6410844) Bugtraq: 6195248,6410592,6482211,6566033,6513116,6565951 Issue: 2314,2313,2312,2290,2289 Review: RD JG-2007-06-25-0: Bugfix: qstat -s hX filtering is broken on darwin-ppc Bugster: 6525883 Issue: 2191 Review: RD GH-2007-06-22-0: Bugfix: NFS write error on N1GE trace file Bugster: 6541085 Issue: 2304 Review: AH, HP RC-2007-06-22-0: Bugfix: qhost -xml lacks '>' (small fix) Issue: 2308 Review: RC (patch submitted by Rayson) RD-2007-06-22-0: Bugfix: - advance reservation did not honor master queue - fixed qrdel race condition with recently deleted job Enhancem: - implemented qacct -ar Review: AA MD-2007-06-22-0: Bugfix: merged from V61_BRANCH (MD-2007-06-22) Review: JG CR-2007-06-22-0: Cleanup: code cleanup in function sge_del_userprj Review: RD AA-2007-06-21-0: Cleanup: missing copyrights in Xpm files JG-2007-06-21-0: Bugfix: qquota filtering broken for pe and project Bugster: 6539199 Issue: 2300 Review: RD MP-2007-06-20-0: Bugfix: inst_sge -ux uninstallation of exechost tries to delete local spooldir, even if it isn't configured. Bugster: 6528949 Issue: 2302 Review: MD JG-2007-06-20-0: Bugfix: sgeremoterun didn't work with Grid Engine >= 6.0 Bugtraq: 6536039 Review: MD RD-2007-06-20-0: Enhancem.: - Added return value 25 for qrsub - Enhanced man pages for advance reservations Review: PJ JG-2007-06-19-0: Bugfix: Grid Engine commands failed, when the primary group of the user was too big in groups file/map. Issue: 2249 Bugtraq: 6568575 Review: RD, AA PJ-2007-06-18-1: Enhancem.: qrsh, qsh, qlogin support the -ar option Review: RD PJ-2007-06-18-0: Enhancem: - Manpages for Advance Reservations Review: DT AA-2007-06-18-0: Bugfix: csp mode /etc/services fix Bugster: 6548455 Issue: 2296 Review: JG PJ-2007-06-15-0: Enhancem: - Manpages for Advance Reservations Review: RD AH-2007-06-15-1: Bugfix: dtrace monitor.sh script should use SGE_CELL variable instead of own command line parameter -cell Bugster: 6569570 Issue: 2294 Review: CR AH-2007-06-15-0: Bugfix: - Integrated resource quota and queue matching code for sequential jobs - Use resource quota result caching also for parallel jobs - Added quick exit for parallel matching code when just a single slot is requested - Duplicate scheduling info messages for reservation jobs - Resource reservation broken for sequential jobs depending on RQS specified for subset of queues only - Fixed issue with wildcard PE requests being used during parallel matching for a specific PE Bugster: 6564461,6569088 Issue: 2274,2293 Review: RD AA-2007-06-14-0: Enhancem: - jgdi_qhost and qhost xml job handling - fixed jgdi tests - CULL_CONFIGURE -> CULL_JGDI_CONF - cleanup and split of jgdi_wrapper.c - handling of CULL_JGDI_RO fields - qmon looses sharetree changes fix Bugster: 6280747,6548455,6560442,6546807 Issue: 2265,2296,2232 Review: RD JG-2007-06-13-0: Bugfix: util/arch had problem with libc version containing comma (e.g. on RH5, Ubuntu 7.04). Bugster: 6568712 Issue: 2241 Review: RD RD-2007-06-11-0: Bugfix: fixed some qrdel and qrstat bugs Review: PJ RD-2007-06-04-0: Bugfix: - reenabled AR related job verification Cleanup: - moved schedule_based decision into sge_assignment structure Review: AA PJ-2007-06-01-1: Bugfix: qalter -ar cause ar is deleted change ar id on runnig jobs Review: JG AH-2007-06-01-0: Bugfix: sge_schedd deadlock upon schedd_job_info job_list being enabled Bugster: 6564503 Issue: 2275 Review: CR RD-2007-05-31-1: Enhancem: added arusers userset Review: JG RD-2007-05-31-0: Bugfix: - fixed core dump with calendar modification - fixed spooling issue with ARs Review: JG JG-2007-05-30-0: Bugfix: Fixed JG-2007-05-29-2. Review: AH RD-2007-05-29-2: CVS Created V61AR_BRANCH ------------------------------------------------ RD-2007-05-29-1: Bugfix: - reverted AR related job verification - fixed scheduler bug with pe reservation Review: AA JG-2007-05-29-2: Bugfix: Completed checkin JG-2007-05-29-0 (owner). Issue: 2229 Bugster: 6544869 Review: AH AA-2007-05-29-1: Bugfix: compiler warnings AA-2007-05-29-0: Enhancem: - Advance Reservation qmon changes - various qmon bug fixes: - split of qrsub sge_parse_qrsub into clients/common/sge_qrsub.c - reserved_slots -> resv_slots for consistency - jgdi delete job crashes master Issue: 1729, 2262, 2263, 2260, 2261, 747, 2264, 2301 Bugster: 6327539, 6538740, 4818801, 4742097, 6553066, 6291044 6555953, 6551121 Review: RD AH-2007-05-29-0: Bugfix: - forgot error exit with AH-2007-05-25-2 memory leak fix - forgot to consider PE resource quota scopes with sequential job optimization AH-2007-05-03-0 Bugtraq: 6562190, 6551568 Issue: 2187, 2238 Review: RH JG-2007-05-29-1: Bugfix: Fixed messages macros. Review: RD RD-2007-05-29-0: Enhancem: - added qrdel -f implementation - fixed some advance reservation issues Review: JG JG-2007-05-29-0: Bugfix: UNKNOWN group/owner in accounting. Issue: 2229 Bugster: 6544869 Review: AH RN-2007-05-28-3: Bugfix: Wrong exit code with qconf -sds. Bugtraq: 6522273 Review: JG RN-2007-05-28-2: Enhancem.: Numbers in error mail too large Bugtraq: 6367642 Issue: 804 Review: JG RN-2007-05-28-1: Enhancem.: qsub -terse to just output job id Bugtraq: 6355875 Issue: 916 Review: JG RN-2007-05-28-0: Bugfix: Still used usersets can be removed under certain conditions Bugtraq: 6422335 Issue: 2050 Review: JG PJ-2007-05-25-1: Bugfix: Joined check for AR jobs that requests more than reserved resources with -w e switch Review: RD LP-2007-05-25-0: Enhancem.: Using aimk debug flag when building libjgdi if set. Enhancem.: Improved and modified GEObjectEditor code for new JGDI functions Enhancem.: JGDIShell improved Enhancem.: Introduced cluster.config.file.location property to override cluster config Review: AA JG-2007-05-25-0: Bugfix: Fixed checkin "Queue request -q from sge_request can't be overridden through command line" Review: RD AH-2007-05-25-2: Bugfix: - PE scope of resource quotas is not considered when -pe wildcard requests are used - memory leak in sge_schedd Bugtraq: 6559409, 6562190 Issue: 2254, 2187 Review: JG PJ-2007-05-25-0: Bugfix: Added check for AR jobs that requests more than reserved resources Added field error_handling to qrstat -ar (-xml) Review: RD RD-2007-05-25-0: Bugfix: qmaster may crash with projects or usersets used in RQS Bugster: 6558006 Issue: 2250 Review: JG OJ-2007-05-23-4: Bugfix: Fixed man page error: man page sgepasswd(5) should be removed Added explanations in sge_shepherd(8), sge_execd(8) regarding sgepasswd usage. Bugster: 6446203 Review: MP OJ-2007-05-23-3: Bugfix: Fixed man page error: reserved usage not explained in sge_conf.5 Bugster: 6447133 Review: MP OJ-2007-05-23-2: Bugfix: Fixed man page error: shell_start_mode should be documentated to be only used for batch jobs Bugster: 6428495 Review: MP OJ-2007-05-23-1: Bugfix: Fixed man page error: reporting lacks information about sharelog records Bugster: 6317563 Review: MP OJ-2007-05-23-0: Bugfix: Fixed man page error: qrsh man page doesn't explain which options don't work with interactive job Bugster: 6364661 Review: MP JG-2007-05-23-0: Bugfix: Fixed resource debitation when modifying exec hosts. Fixed a memory leak in scheduler. Review: AH Bugfix: Fixed queue status string generation. Review: RD JG-2007-05-22-0: Bugfix: Queue status in reporting file is missing. Bugster: 5081743 Issue: 2258 Review: JG RD-2007-05-21-0: Bugfix: - Fixed return value for qselect and some qconf switches - fixed bug with host complexes consistency check Cleanup: Code uses now GDI request bitmask Review: AA LP-2007-05-18-0: Bugfix: Calling JGDI getQueueInstanceSummary results in a memory leak Bugtraq: 6559385 Review: AA AH-2007-05-14-0: Bugfix: Memory leak in libdrmaa.so Bugster: 6551121 Review: LP AH-2007-05-11-1: Bugfix: Fixed resource reservation broken by AH-2007-05-03-0 AH-2007-05-11-0: Bugfix: Add -u to scheduler category only if there is a resource quota for the user Bugster: 6554313 Issue: 2248 Review: TBD MP-2007-05-11-0 Bugfix: Queue request -q from sge_request can't be overridden through command line Bugster: 6328064 Review: JG EB-2007-05-10-1: Bugfix: xacl output was missing in qrstat output Review: TBD AA-2007-05-10-0: Bugfix: qmon crashes when displaying about dialog Bugster: 6555744 Issue: 2243 Review: RD EB-2007-05-10-0: Enhancem.: Handle complex value modifications in queue and host correctly when AR are present Review: RD RD-2007-05-09-1: Enhancem.: Added check if userset mod invalidates AR Cleanup: moved userset code into generic gdi framework Review: AA JG-2007-05-09-0: Bugfix: job_get_id_string was not MT save Bugster: 6476263 Review: RD RD-2007-05-09-0: Bugfix: Fixed abort in several clients due to AH-2007-05-03-0 Review: AA JG-2007-05-08-1: Bugfix: Remove util/sge_log_tee (moved it to source/scripts) Bugster: 6426331 Review: RD PJ-2007-05-08-1: Bugfix: Forgotten message breaks the readability PJ-2007-05-08-0: Enhancem.: qrdel -u Support for testsuite test Review: RD JG-2007-05-08-0: Bugfix: inst_sge -m failed, when called as non root user and the USER environment variable was not set. Bugster: 6536426 Issue: 2219 Review: MD AH-2007-05-07-0: Bugfix: Use of hostgroups in resource quotas is less performant than the full list of hosts Bugster: 6542137 Issue: 2239 Review: OJ AH-2007-05-03-0: Bugfix: - Need faster resource quota matchmaking and more concise job info messages - qstat -j lacks resource quota info messages in case of "incomplete" resource quotas Bugster: 6551568, 6550718 Issue: 2238 Review: OJ RD-2007-05-02-0: Bugfix: - implemented AR error state handling - implemented AR mail sending - fixed qdel on job in deletion state Issue: 1860 Bugster: 6345522 Review: EB EB-2007-05-02-0: Enhancem.: changes for qstat and advance reservation Review: RD RD-2007-04-27-0: Bugfix: fixed broken distinst Review: JG PJ-2007-04-27-0: Bugfix: qrsub -d INFINITY Review: RD JG-2007-04-26-0: Enhancem.: New architecture strings ulx-* and usol-* for older, no longer officially supported OS versions. Some cleanup of distinst and sge_arch.c. qmake and qtcsh for lx26-amd64. Merged JG-2007-04-19-0 and JG-2007-04-24-0 from V61_BRANCH. Review: RD PJ-2007-04-25-0: Bugfix: qrstat_report_handler_xml Update the AR TODO LIST Review: RD PJ-2007-04-24-2: Bugfix: compile error, qrsub xacl_list fix PJ-2007-05-08-0: Enhancem.: qrdel -u Support for testsuite test Review: RD PJ-2007-04-24-1: Bugfix: qrsub xacl_list fix PJ-2007-04-24-0: Bugfix: Classic spooling for AR, added spooling for granted slots Minor bugfix and changes Bugfix: qrsub -u RD-2007-04-23-0: Enhancem: Added consistency checks for calendar and advance reservations Review: EB AH-2007-04-20-0: Cleanup: Removed unused ticket scheduler code and data fields Review: JG RD-2007-04-19-1: Bugfix: Fixed DRMAA Javadoc Warnings Review: RH EB-2007-04-19-0: Enhancem.: - reject pe_list and ckpt_list changes of cqueues if they invalidate advance reservations - reject slots modification in pe if they invalidate - advance reservations Review: AH GH-2007-04-18-0: Bugfix: Wrong path to spooled parallel jobs with using classic spooling RD-2007-04-19-0: Enhancem.: - Added advance reseservation scheduling code - added qmaster startup AR initialization code - added removing of AR jobs at AR deletion - added reporting entry at qrdel AR deletion Review: EB JG-2007-04-18-1: Enhancem.: Cleanup / higher precision for calculation of the cpu load value on HPUX. Review: AH JG-2007-04-18-0: Bugfix: Modifying a rqs with invalid syntax resulted in its deletion. Bugster: 6528950 Issue: 2233 Review: AP, MD, MP, OJ GH-2007-04-18-0: Bugfix: Wrong path to spooled parallel jobs with using classic spooling Bugtraq: 6461308 Review: AH RH-2007-04-18-0: Enhancem.: xml schema for qhost -xml output added Bugfix: -xml switch in qhost man page documented Bugfix: qstat -F -xml does not show resources Bugster: 6546802 Issue: 2231 Review: RD DT-2007-04-16-0: WWW: o New Howto link AH-2007-04-16-0: Bugfix: Fix AH-2007-03-29-2 plus further optimizations with schedulers quota verification time: Detect global, cluster queue, or host related quota saturation and exploit such cases unblushingly. Optimization can be applied only if the first quota rule in a rule set is satured. Removed unused code. Bugster: 6539215 Review: RD PJ-2007-04-13-0: Bugfix: Classic spooling for AR Minor bugfix and changes Enhancm: qalter -ar option Bugfix: ~/.sge_ar_request was in error Review: RD EB-2007-04-12-1: Bugfix.: - fixed code for classic spooling (pe_range) - instead of event/state numbers stere are letters/strings written EB-2007-04-12-0: Enhancem.: - implemented qrstat - added man pages Review: RD EB-2007-04-10-0: Bugfix: - fixed bug in new_job reporting (incorrect format string) PJ-2007-04-05-0: Bugfix: - fixed bug qrsub -M list EB-2007-04-05-0: Enhancm: - removed ar_id from new_job entry in reporting file - now ar_acct entries are created on termination of an ar in reporting file - ar related reporting was only written when it is enabled Review: RD PJ-2007-04-05-0: Enhancm: - Disabled some options according the specification - Updated ussage of the qrsub -help Review: RD MP-2007-04-05-0 Bugfix: Root user on unix unable to submit job to windows execution daemon. Bugster: 6513597 Review: JG RD-2007-04-04-2: Bugfix: - fixed qsub -ar check if no hr_t was selected - added ckpt_name for AR reserving decision - added AR TODO list Review: HP RD-2007-04-04-1: Bugfix: fixed broken build on lx24-ia64 RD-2007-04-04-0: Enhancm: - Added Advance Reservation reserving code - fixed consumable resources testsuite test - enhanced test_resource_utilization Review: EB RC-2007-04-03-0: Enhancm: A small tight SGE-SSH integration change Review: pending JG-2007-04-02-0: Bugfix: Hybrid user/project sharetree was broken with array jobs. Bugster: 6538293 Issue: 2222 Review: AH PJ-2007-03-30-0: Enhancm: added and tested qrsub all options as are defined in FSD 6.2 AR document Review: RD DT-2007-03-28-0: WWW: o Added link to the YouTube video to the howtos page. AH-2007-03-29-2: Bugfix: Quota verification time of scheduler may not grow with the number of queues if all jobs go in single queue Bugster: 6539215 Review: RD AA-2007-03-29-1: Enhancem.: insure support for jgdi Job testvalues Review: LP AH-2007-03-29-0: Bugfix: Resource quotas broken after qmaster restart Bugster: 6539792 Issue: 2224 Review: RD AA-2007-03-28-2: Bugfix: reverted jgdi_qhost to previous version DT-2007-03-28-0: WWW: o Added link to the YouTube video to the main page. AA-2007-03-28-1: Bugfix: - template file comments break build on Aix AA-2007-03-28-0: Enhancem.: - jgdi maplist/map handling - CULL_CONFIGURE adjustments in testvalues/*/*.xml - fix of jgdi tests - jgdi event fixes - sge_gdi_del_job() master fix for invalid idep - cleanup: author tags - jgdi advance reservation test fixes Review: RH PJ-2007-03-28-0: Cleanup: all attr_mod functions are now located in sge_utility_qmaster.h Review: AH EB-2007-03-26-0: Enhacem: - added reporting funtionality for AR. - added first implementation of classic spooling for AR. - changed accounting file format and qacct (added arid). - changed "acct" entry in reporting file (added arid) - changed man page - added comments Review: RD PLEASE NOTE: Reporting and accounting file format changed. This might have influence on scripts/applications which reuse that information (e.g. dbwriter) HP-2007-03-26-1: Bugfix: Fixed broken build for sol-sparc64 and -no-java Changed: aimk Review: AA RH-2007-03-26-1: Cleanup: jgdi: EventType has to be serializable Review: AA RH-2007-03-26-0: Cleanup: cleanup in jgdi code - missing "implements Serializable" added - author tags deleted Review: AA AH-2007-02-23-0: Enhancem. Enhanced analyze.rb script - extended meaning of -t to all views - add maxvmem and maxrss to all views MP-2007-03-23-0: Bugfix: Add installation of SGE_Helper_Service to auto installation. Bugster: 6440226 Issue: 2220 Bugfix: Binary check in inst_sge is wrong. Windows does not require sgepasswd, but unix should. Bugster: 6521802 Issue: 2221 Docs: Modified man page to reflected new qmaster parameter MAX_JOB_DELETION_TIME Review: JG GH-2007-03-23-0 Bugfix: problems with sge_qstat(5) file Bugtraq: 6452471 Issue: 2088 Review: AH RD-2007-03-22-0: Bugfix: Corrected and validated qquota schema file Review: RH PJ-2007-03-21-0: Enhancm: added job -ar related checks Review: RD JG-2007-03-21-0: Cleanup: Fixed message macros. Review: CR LP-2007-03-20-0: Bugfix: Removed unused import in EditorParser.java that prevents compilation on Java 1.6 Review: PJ LP-2007-03-19-2: Enhanc: Introducing GEObjectEditor class for simple GEObject manipulation based on text Qconf options implemented for User, Project, PE, ClusterQueue, Calendar, Checkpoint, Exec host, Host group and RQS (some issues exist). Minor improvements/fixed to JGDI code. Changed: JGDI Review: AA LP-2007-03-19-1: Bugfix: Fixed incorrect use of DEXIT in jgdi_common.c Review: PJ LP-2007-03-19-0: Bugfix: JGDI does not accept null values in certain cases anymore Bugtraq: 6535877 Review: AA AH-2007-03-19-0: Cleanup: Removed unused variables and parameters from ticket scheduler code. Review: AA AH-2007-03-16-1: Enhanc: Added tracing of job script spooling operations to Dtrace master monitor. Changed: qmaster Review: JG RD-2007-03-16-0: Bugfix: - added missing check for JB_verify_suitable_queues - fixed several compile errors Review: AH PJ-2007-03-16-0: Enhancm: - added qrsub command line options - added qmaster advance reservation verification - added qsub -ar option Review: RD AA-2007-03-16-0: Bugfix: o qmon advance reservation qrstat, qrsub dialogs prelim version available via task menu o sge_c_gdi.c return AR_id o resource quotas CULL_CONFIGURE flag o lMt2Str helper func Review: RD RH-2007-03-15-0: Bugfix: o jgdi: interface for creating an event client changed (was not threadsafe) o SEGV in sge_log.c Review: AA AH-2007-03-15-0: Enhanc: Added SIMULATE_EXECDS to qmaster_params to to bypass execd communication in qmaster for throughput tests. Changed: qmaster JG-2007-03-14-0: Bugfix: Fixed utilbin/hp11-64/rshd. Review: RD AH-2007-03-13-1: Docs: Mention need to link sge_pe(5) with a queue with the 'pe_list. Changed: sge_pe(5) JG-2007-03-13-0: Bugfix: qconf -mqrs modified resource quota, even if the editor was left without saving. Bugster: 6533754 Issue: 2207 Review: AA MP-2007-03-13-0: Enhanc: created new qmaster parameter for max job deletion time. Changed: sge_conf Review: JG JG-2007-03-07-1: Cleanup: Fixed typos, added documentation, cleanup. Review: AA AA-2007-03-07-0: Bugfix: qstat -r -xml is not working Bugster: 6531921 Issue: 2206 Review: JG Changed: qstat JG-2007-03-07-0: Bugfix: Modifying a resource quota set while there are jobs running could crash qmaster. Bugster: 6530335 Issue: 2202 Review: AA MP-2007-03-07-0: Bugfix: scalability issue with qdel and very large array jobs. Changed: qmaster Bugster: 6288953 Issue: 752 Review: JG AA-2007-03-06-0: Bugfix: qstat -xml does not show pending/zombie jobs Bugtraq: 6531317 Issue: 2204 Review: JG Changed: qstat AH-2007-03-01-1: Docs: Removed wrong and confusing example in host_conf(5). Issue: 2200 Review: Reuti Changed: Man pages AA-2007-02-27-0: Bugfix: sge_ca script on no root access fs Bugtraq: 6528808 Issue: 2198 Changed: sge_ca Review: CR MD-2007-02-26-0: Bugfix: Auto installation option failed to save the install log Changed: inst_sge, install_modules Bugtraq: 6472614 Issue: 2196 Review: HP MP-2007-02-22-0: Bugfix: on Windows, execd doesn't provide useful error messages when SSL keys broken Bugtraq: 6395075 Changed: utilbin/sge_passwd.c utilbin/msg_utilbin.h Review: HP LP-2007-02-20-0: Bugfix: JGDI crashes JVM when null is passed to JNI GetStringUTFChars function Bugtraq: 6525497 Review: RH JG-2007-02-16-0: Bugfix: Fixed non initialized pointer access in qacct. Merge of V61_BRANCH, JG-2007-02-15-0. Bugster: 6525917 Issue: 2192 Review: RD MD-2007-02-14-0: Enhancem: o CSP mode auto installation: certificates are not copied to submit hosts Bugfix: o 64 bit solaris BDB rpc server broken Changed: inst_sge, install_modules Bugtraq: 6421113, 6291021 Issue: 2188, 1445 Review: JG AH-2007-02-14-2: Bugfix: Added 'dtrace' directory to common list in scripts/mk_dist. Review: RD RD-2007-02-14-1: Cleanup: removed postgresql specific scripts from installation Review: JG Issue: 1105 AH-2007-02-14-0: Docs: - Corrected queue_conf(5) where qconf(1) option "-sds" was wrongly refered "-sick". - need to define continuation character behaviour with configuration files - describe useful characters for every parameter with qconf file formats Bugtraq: 6520257, 6521285 Changed: Man pages Review: RD RD-2007-02-14-0: Bugfix: fixed qmon lib error on darwin Review: JG DT-2007-02-13-0: Docs: o Corrected man pages from PJ-2007-02-09-0 Changed: man/man1/sge_types.1, man/man5/complexes.5 RD-2007-02-12-0: Enhancem: - Added advance reservation object to qmaster - added rudimental qrsub/qrdel/qrstat clients Review: AA RD-2007-02-09-1: CVS Created V61_BRANCH ---------------------------------------------- PJ-2007-02-09-0: Bugfix: Manpage complex and sge_types correction. Review: DT RD-2007-02-09-0: Bugfix: - corrected qdel help output - corrected resource quota man pages - fixed distinst for sgepasswd on win32-x86 Review: AA HP-2007-02-09-0: Bugfix: roll back of MP-2007-02-08-0 to RD-2007-02-01-0 to fix broken build. Changed: utilbin/sge_passwd.c utilbin/msg_utilbin.h MP-2007-02-08-0: Bugfix: on Windows, execd doesn't provide useful error messages when SSL keys broken Bugtraq: 6395075 Changed: utilbin/sge_passwd.c utilbin/msg_utilbin.h Review: HP AA-2007-02-08-0: Bugfix: qmon crash in cluster configuration dialog when modifying a host, incomplete merge from 60u10 fix: AA-2006-12-22-0, CR 6353526/IZ 1904 Bugtraq: 6522385 Issue: 2185 Review: RD AA-2007-02-07-0: Enhancem: support of HP1164 Review: JG JG-2007-02-06-0: Bugfix: meaningless error message in clients when reporting_param flush_time is incorrectly set. Bugtraq: 6499217 Issue: 2183 Review: RD RD-2007-02-05-0: Enhancem: added qquota schema file Review: AA MD-2007-02-05-0: Enhancem: o remote installation of execd's need enhancement, rework, cleanup Changed: inst_sge, install_modules Bugtraq: 6513433 Issue: 2182 Review: HP JG-2007-02-05-0: Bugfix: Fixed build of hp11-64. Cleanup. Review: AA PJ-2007-02-02-0: Bugfix: Broken output of job name with 'qsub -N'. Issue: 2061 Bugtraq: 6494390 Review: RD RD-2007-02-02-1: CVS Created V61preview2_TAG --------------------------------------------------- RD-2007-02-02-0: Bugfix: removed hp11-64 from arch and arch.dist RD-2007-02-01-0: Bugfix: - added 32bit libs to sol-sparc64 in distint & mk_dist - added resource_quotas classic spooling to backup - enabled HostTab in qmon AA-2007-02-01-0: Bugfix: Fixed loading of openssl amd libspoolb/c on hp11-64 aimk -javadoc exit dist/util/arch.dist hp11-64 support Review: RD JG-2007-02-01-0: Bugfix: Fixed commlib on hp11-64. Review: RD JG-2007-01-31-0: Bugfix: Fixed initialization of spoolinit. Fixed distinst with static linking of bdb. Fixed broken build on tru64. Review: AA LP-2007-01-31-1: Bugfix: Project man page contains different attribute names. Issue: 2181 Bugtraq: 6518689 Changed: project.5 Review: JG LP-2007-01-31-0: Bugfix: Qconf usage x man page inconsistency Issue: 2180 Bugtraq: 6518684 Changed: msg_common.h Review: JG JG-2007-01-30-2: Bugfix: Fixed broken build on win32-x86. Build depends on both TMPDIR (gcc) and TMP (cl.exe) environment variables. Allow to set TMP via the aimk.site/aimk.private setting WINTEMPDIR. Review: AA, HP JG-2007-01-30-1: Bugfix: Fixed a memory access problem in qrsh_starter. Review: RD JG-2007-01-30-0: Bugfix: Fixed qconf manpage (-aconf/dconf/mconf/sconf switches) Issue: 2105 Review: AH HP-2007-01-29-1: Enhancem.: On SOLARIS64, if no target and no "-only-*" parameter is specified, SOLARIS shared libraries are also built now. Changed: aimk Review: JG AA-2007-01-29-0: Bugfix: Fixed access to already freed memory in cl_get_com_handle() added cl_thread_cleanup_global_thread_config_key() Bugfix: 6518607 Issue: 2178 Review: CR JG-2007-01-29-2: Bugfix: Fixed access to already freed memory in sge_execd and sge_shadowd. Review: AA JG-2007-01-29-1: Enhancem.: Added module test to verify if build options contain the necessary options and defines for multi threading. Review: RD JG-2007-01-29-0: Bugfix: Fixed memory access errors. Review: RD AA-2007-01-26-0: Enhancem: o build support for aimk -32/-64 for jgdi for hp11/hp11-64 and sol-sparc/sol-sparc64 Review: RD JG-2007-01-26-0: Bugfix: Fixed a typo in complex.5 manpage. Issue: 1946 EB-2007-01-25-0: Bugfix: qhost -l selected unsufficient number of hosts Review: JG JG-2007-01-25-0: Bugfix: qhost didn't output error message in certain error situations (e.g. no bootstrap file exists). Review: AA MD-2007-01-24-0: Enhancem.: added LicenseAgreement to master installation Review: HP DT-2007-01-23-1: Docs: o Updated the DRMAA howto page and classes for 1.0. Changed: www, Java howto examples PJ-2007-01-23-0: RFE: Evaluation Expression Support. Test suite fix PJ-2007-01-19-0, detected in testsuite/tests Review: AH AA-2007-01-23-0: Enhancem: o problem in qsh, qrsh due to fix AA-2007-01-19-0 detected in testsuite/tests Bugtraq: 6515122 Issue: 2166 Review: RD DT-2007-01-23-0: Cleanup: o Added missing mk_dist support for DRMAA 1.0 Changed: mk_dist RD-2007-01-23-1: Cleanup: added missing checkin for RD-2007-01-22-0 Review: AA RD-2007-01-23-0: Doc: unified time definition for accounting and reporting man page Review: AA DT-2007-01-22-0: Cleanup: o Replaced distinst -s2 switch with base dir in distinst.site Review: pending(AH) Changed: distinst, distinst.site HP-2007-01-22-0: Cleanup: Replaced CR+LF with CR in num_proc output for HP-2007-01-19-0 Changed: loadcheck.c Review: MD RD-2007-01-22-0: Cleanup: - removed unused clients in definitions - switched qdel_help to sge_help CR-2007-01-22-0: Cleanup: removed unreferenced message macro definitions DT-2007-01-19-1: Enhancem.: o Set target source levels for DRMAA Java language binding. Changed: classes/build.xml Issue: 6403810 DT-2007-01-19-0: Enhancem.: o Upgraded DRMAA Java language binding to 1.0 o Made the DRMAA C binding 1.0 the default - 0.5 Java binding moved to s2 branch Changed: classes, man pages, JAPI, tests, distinst Review: AH + pending HP-2007-01-19-0: Bugfix: loadcheck reports on a hyperthreaded CPU only one processor on Windows Changed: loadcheck.c, utilbin/Makefile, aimk, distinst inst_common.sh, inst_execd.sh Bugster: 6233523 Review: MD PJ-2007-01-19-0: RFE: Evaluation Expression Support. Issue: 56 Bugster: 6205203,5101053,5101735 Review: AH AA-2007-01-19-0: Enhancem: o -wd functionality in qsub,qalter,qsh,qrsh and qmon o sge_event_client2.* -> sge_event_client.* o act_qmaster, bootstrap error messages o mk_dist usage Bugtraq: 6515122 Issue: 2166 Review: RD MD-2007-01-18-0: Bugfix: path to openssl lib in distinst.site was wrong Reveiw: RD Changed: distinst.site GH-2007-01-18-0: Bugfix: problems with sge_qstat(5) file Bugtraq: 6452471 Issue: 2088 Review: AH RD-2007-01-17-0: Bugfix: qmaster core dump with usersets referenced in RQS unix groups are not considered by RQS Issue: 2161, 2162 Bugster: 6513944, 6513967 Review: JG CR-2007-01-17-0: Bugfix: scheduler now writes pid file in daemonize phase Bugtraq: 6516288 Issue: 2171 Review: JG MD-2007-01-17-0: Bugfix: merged from V60s2_BRANCH RD-2007-01-12-0: Doc: added Advance Reservation Specification Document AH-2007-01-11-0: Enhancem: Enabled distinst code that assumes $ORIGIN relative RUNPATH is set with OpenSSL and BDB libraries and binaries. If no such libraries/binaries are available distinst option -shlibpath still can be used to cause old LD_LIBRARY_PATH behaviour be effective. (see AH-2006-10-12-0) Bugster: 6456290 Issue: 2090 Review: RD EB-2007-01-10-0: Bugfix: qmaster dies when modifying slots value for queue domain when queuename is missing Bugtraq: 6509684 Issue: 2158 Review: JG JG-2007-01-10-0: Bugfix: o spooledit didn't dump QINSTANCE objects of version 6.0 o spooledit didn't dump USERSET objects Bugster: 6511171 Issue: 2157 Review: RD RD-2007-01-09-0: Bugfix: Default requests for complexes not honored by resource quotas Bugtraq: 6510635 Issue: 2146 Review: JG AA-2007-01-08-0: Enhancem: configurable flag for jgdi objects (maybe incomplete) Review: RH JG-2007-01-08-0: Bugfix: Backup script failed in case of inconsistent hostname resolving. Review: MD Cleanup: Fixed typo. DT-2007-01-05-0: Docs: o Reviewed all man page changes since u4 o Corrected language errors Changed: Man pages OJ-2007-01-03-0: Bugfix - qconf -aq help did not display correctly Modified the string to read now [-aq [queue_name]] in V60s2_BRANCH Issue: 1885 Review: DT HP-2007-03-01-1: Bugfix: fixed broken build on ALPHA Changed: aimk HP-2007-03-01-0: Bugfix: wrong qlogin_daemon or rlogin_daemon in host conf doesn't set host and job into error state Changed: execution_states.c, reaper_execd.c, builtin_starter.c sge_uidgid.c Bugtraq: 6349814 Issue: 1914 Review: MD MD-2007-01-03-0: Bugfix: autoinstall fix, EXECD_HOST_LIST check failed merges from V60s2_BRANCH Review: JG Changed: install_modules RC-2006-12-28-0: Enhancem.: built-in support for getting load info for AIX Changed: libuti Review: HP RC-2006-12-26-0: Bugfix: fixed a typo Changed: doc/devel/rfe/ResourceQuotaSpecification.html OJ-2006-12-22-X: Bugfix: - qconf -aq help did not display correctly Modified the string to read now [-aq [queue_name]] Issue: 1885 Review: DT RD-2006-12-22-0: Bugfix: - qconf -Arqs added invalid RQS - removed some unnecessary memset() calls - Resource quota set breaks the scheduler behavior with multiple queue instances on exec hosts - load formula does not recognized float as weighting factor - enhanced load formula/dynamic RQS limit syntax Bugster: 6507572, 6506129, 6507576, 6507579 Issue: 2150, 2149, 2151, 2152 Review: JG AA-2006-12-22-0: Bugfix: Merge of V60u10 fixes and Hintak add-ons o reprioritize field in qmon cluster config missing (CR 6353526/IZ 1904) o host already exists when modifying cluster settings (CR 6494508/IZ 2080) o qmon should refresh job list when switching tabs (IZ 2111) o qmon: missing dependency on libevc (IZ 2118) o qmon: QIM task error message not displayed in "Why ?" (IZ 2119) o qmon: row numbers of job tabs keep on increasing (IZ 2120) o qmon: adding qhost-like load listing (IZ 2126) o qmon job control: display wider default columns (CR 6506677/IZ 721/IZ 2121) o Sorting by different fields (CR 6506637 / IZ 571) o qmon_howto addition of qhost folder and job control column size descr. Review: RD,AH JG-2006-12-22-0: Bugfix: Fixed broken build introduced by JG-2006-12-20-0. Review: RD CR-2006-12-21-0: Bugfix: qsh does not work if XAUTHORITY is set in root environment BT: 6482762 Issue: 2128 Review: RD JG-2006-12-20-0: Bugfix: sge_shepherd dumped core on linux amd64 for qrsh jobs with very long commandline (> 10k). Bugster: 6506115 Issue: 2141 Bugfix: distinst failed, when berkeleydb is statically linked (no libdb.so available). Bugfix: A part of the java build didn't respect setting of the maxwait property. Doc: Updated README for build of Berkeley DB. Review: RD EB-2006-12-21-1: Bugfix: qselect matches wrong resource which have been overridden at lower level Bugtraq: 6475272 Issue: 2145 Review: RD EB-2006-12-21-0: Bugfix: ignore_fqdn is broken for the local configuration Bugtraq: 6488244 Issue: 2144 Review: RD CR-2006-12-20-0: Bugfix: fixed aimk for windows compile Review: MD RD-2006-12-18-0: Cleanup: Changed resource quota expanded list format Review: JG HP-2006-12-15-0: Cleanup: Removed special target for Windows authuser.exe Changed: aimk, utilbin/Makefile RD-2006-12-15-0: Bugfix: - Fixed qmaster crash with resource quotas - corrected qconf man page Bugster: 6506115 Issue: 2135 Review: JG RH-2006-12-15-1: Bugfix: Clean target deletes java parts only if executed on java build hosts Changed: jgdi, juti and drmaa build.xml Review: AA HP-2006-12-14-0: Bugfix: Only the execd_params value enable_windowmacc now desides if passwords from sgepasswd are to be used for switching the user, no longer if the user is a Windows domain user, too. Changed: libs/uti/sge_uidgid.c, libs/uti/sge_uidgid.h Bugster: 6504566 Review: MD RH-2006-12-14-1: Bugfix: ant clean targets failed due to nfs problems. Workaround: directories are deleted twice with a sleep between the first and the second delete Review: AA RH-2006-12-14-0: Cleanup: juti: javadoc warning solved, author tags removed Review: RD RD-2006-12-13-0: CVS Created V61preview_TAG -------------------------------------------------- JG-2006-12-13-0: Cleanup: Removed testsuite - it has been moved to the testsuite subproject. HP-2006-12-12-0: Bugfix: Merged missing parts from checkin HP-2006-12-01-1 Removed debugging code Added trace info for job_start_remote Changed: shepherd RD-2006-12-11-0: Enhancem.: Added arco switch to distinst & mk_dist Changed: scripts/mk_dist scripts/distinst Review: AA HP-2006-12-11-0: Enhancem.: Added Windows version of authuser binary Changed: utilbin/Makefile AH-2006-12-11-0: Enhancem.: Determine qmaster_spool_dir based on bootstrap file in Dtrace monitor AH-2006-12-08-1: Enhancem.: Added probe verification option to Dtrace monitor AH-2006-12-08-0: Enhancem.: Added Dtrace monitor to distribution Review: RD HP-2006-12-08-0: Enhancem.: Code cleanup and improved documentation for checkin MP-2006-12-06-0 Changed: shepherd/builtin_starter.c, libs/uti/sge_uidgid.c RD-2006-12-08-0: Docs: Corrected examples in Resource Quota Spec Review: JG RD-2006-12-07-0: Docs: Clearified mem description in accounting.5 Review: AH MP-2006-12-07-1: Bugfix: Adding missing error handing for calls to sge_set_uid_gid_addgrp() function Changed: daemons/shepherd/builtin_starter.c Review: HP MP-2006-12-07-0: Bugfix: error handling didn't account for value of 1 Changed: source/daemons/shepherd/builtin_starter.c Review: HP RD-2006-12-06-0: Enhancem: PDC for DARWIN improved qquota help output Changed: execd, shepherd Issue: 2130 Bugtraq: 6501447 Review: JG MP-2006-12-06-0: Bugfix: fixed shepherd error state when there is a problem with the sgepasswd file only when the file is missing should the whole shepherd go into error state, otherwise the job should go to error state. Changed: source/common/execution_states.h source/common/execution_states.c source/daemons/execd/reaper_execd.c source/daemons/shepherd/builtin_starter.c source/libs/uti/sge_uidgid.c Bugster: 6347256 Review: HP AH-2006-12-05-0: Bugfix: sge_conf(5) setting 'max_u_jobs' broken, if BDB spooling is used Bugster: 6500359 Issue: 2129 Review: JG AH-2006-12-04-0: Bugfix: qmaster performance gets throttled if qsub -sync y is used when many jobs are in the system (fix test_drmaa ST_RECOVERABLE_SESSION) Bugster: 6497575 Issue: 2125 Review: RD CR-2006-12-04-0: Bugfix: hostname resolving should not be case sensitive Issue: 1905 BT: 6353558 Review: RD HP-2006-12-01-1: Enhancem.: Enhanced error handling for Windows GUI jobs Changed: shepherd, execd Review: MD HP-2006-12-01-0: Bugfix: o replaced usage of UID 0 and "root" with calls to platform independent functions o map gdi request user name "Administrator" to "root" on Windows, so the QMaster accepts request to be from privileged user. Changed: libs/gdi/sge_gdi2.c libs/gdi/sge_security.c libs/uti/sge_uidgid.c libs/uti/sge_uidgid.h Review: AA JG-2006-11-30-0: Enhancem.: NetBSD port (merge from V60_s2_BRANCH). Removed -Wno-strict-aliasing in Linux builds. Review: RD AA-2006-11-29-0: Bugfix: sge.root property assertion removed in libs/jgdi/build.xml AH-2006-11-29-0: Bugfix: qsub -sync y is used when many jobs are in the system Bugster: 6497575 Issue: 2125 BD-2006-11-28-1: Bugfix: Don't include nonexistant crypt.h on FreeBSD (as on Darwin) Changed: source/utilbin/authuser.c BD-2006-11-28-0: Enhancem: PDC for FreeBSD Changed : execd, shepherd Review: RC RD-2006-11-28-0: Bugfix: - recheckin of RD-2006-03-22-0 changes were lost due to gdi2 merge - fixed some memory leaks in qhost Changed: qhost, qstat Review: JG HP-2006-11-28-0: Bugfix: fixed broken builds on several architectures moved variable declarations to top of function bodies, cleaned up code. Changed: source/libs/uti/sge_uidgid.c source/libs/uti/sge_uidgid.h HP-2006-11-27-0: Bugfix: fixed memory leak in sgepasswd Changed: source/libs/uti/sge_uidgid.c RM-2006-11-24-0: Bugfix: o fixed sgepasswd segmentatation fault that occured when sgepasswd was called with corrupted sgepasswd file Changed: source/libs/uti/sge_uidgid.h source/libs/uti/sge_uidgid.c source/utilbin/msg_utibin.h source/utilbin/sge_passwd.c Bugster: 6472859 Review: HP CR-2006-11-24-0: Cleanup: o removed needless execd cache for commlib messages o using free in FREE macro (sge_free() needless here) o using FREE macro in clear_packbuffer() Changed: execd, culllib, utilib Review: RD CR-2006-11-23-1: Cleanup: qmaster heartbeat file writing differs from V60s2_BRANCH (seems to be a forgotten merge) Review: RD CR-2006-11-23-0: Bugfix: Fixed wrong shepherd signal handler initalization qdel of just submitted job (merged from s2 branch) Issue: 126 Bugster: 6287963 Changed: qsh, shepherd, utilib Review: AA CR-2006-11-23-0: Enhancem.: new paramater "maxwait" for java build timeout Changed: source/classes/build.xml source/libs/jgdi/build.xml source/libs/juti/build.xml source/build.properties Review: AA HP-2006-11-23-0: Bugfix: Fixed broken build of Win32 binaries Changed: aimk AA-2006-11-22-0: Bugfix: qevent fix for broken testsuite Review: CR JG-2006-11-23-0: Enhancem.: Include java targets in standard build (aimk), -no-java can be used to exclude java targets. Some cleanup. Review: AA CR-2006-11-22-0: Bugfix: - fixed problem in sge_strdup function when an empty string (string with length 0) is duplicated (segmentation fault with empty string) Bugster: 6497217 Issue: 2122 Review: AA, RD RH-2006-11-21-0: Cleanup: - fine tuning of jgdi build process Review: AA RD-2006-11-20-1: Cleanup: - removed sge_bitop.c and sge_crc.c - aimk clean now removes sge_test_mirror - fixed build error for java Review: AA RH-2006-11-20-0: Enhancem.: o jgdi version 0.91 with a threadsafe JGDI connection object Review: AA RD-2006-11-20-0: Bugfix: fixed some issues introduced by resource quota renaming Review: AA RC-2006-11-19-0: Enhancem: fixed some typos Changed: profiling.txt AA-2006-11-17-0: Bugfix: o broken build on solaris86 o moved test_sge_mirror o resource quota renamings Review: RD RD-2006-11-17-0: Cleanup: renamed limitation rule sets to resource quota sets Review: AA RD-2006-11-16-0: Bugfix: - changed default behaviour of qstat -u - fixed some qlimit bugs - backup classic spooling missed lirs directory - fixed qconf -cq return code Bugster: 6382589 Issue: 1976 Review: JG AA-2006-11-15-0: Enhancem: o remove TEST_GDI2 obsolete sections and files o build_testsuite.properties added o show detached settings in jgdi o OutputTable from QinstanceSummaryPrinter extracted Review: RD, RH RH-2006-11-13-0: Bugfix: Broken compile of jgdi on denethor Review: AA JG-2006-11-13-0: Bugfix: Fixed messages file msg_utilbin.h. Review: AA RD-2006-11-10-0: Bugfix: fixed some limitation rule set bugs - qconf -Xattr with limitation_rule set were broken - filter matching on queue level was broken - limit on non consumable were not enforced Review: JG RH-2006-11-09-1: Bugfix: o missing links for jni_libs in target directories o jgdi: run_util.sh supports now darwin-* and hp11* Review: AA RH-2006-11-09-0: Enhancem.: o juti: authuser moved to utilbin o juti: java part for windows user authentication implemented Review: HP HP-2006-11-03-0: Bugfix: o Changed binary detection of installation from authuser to authuser.exe for Windows Changed: inst_common.sh Review: MD AH-2006-11-03-0: Bugfix: Ensure pthread_key_create(&gdi_state_key) be called only once. Review: AA AA-2006-11-02-0: Bugfix: o distinst.site file for openssl-0.9.7l Review: CR RD-2006-11-02-0: Bugfix: o fixed qlimit return value for help output o fixed minor qconf -mlrs issues o fixed reservation bug with limitation rule sets o fixed pe scheduling bug with limitation rule sets Review: JG HP-2006-11-02-0: Enhancem.: added authuser.exe for Windows to distinst script CR-2006-10-31-0: Cleanup: o fixed support for issue 1096 from testsuite (missing code) Review: AA RH-2006-10-31-0: Enhancem. o missing events for jgdi implemented o cleanup for event class generated o SISSL headers added o new test command libs/jgdi/jgdievc (prints all events received over jgdi to stdout) o jgdi support now parallel builds on different hosts ("aimk -java" started on different platforms works). The java build host has to defined in build_private.properties. If this property is not defined, the localhost is used. Review: AA AS-2006-10-30-1: Bugfix: dist/util/sge_request referred to cod_request(5) man page instead of sge_request(5) Bugster: 6325359 AA-2006-10-30-0: Bugfix: o repaired broken build on certain archs through sge_lock fix o fixed error messages for PEM_BYTE code Review: RD RH-2006-10-30-0: Bugfix: o jgdi: QueueInstanceSummary did not work with a queue filter (JB_hard_resource_list has been reduced) Review: AA AA-2006-10-27-0: Bugfix: o changed sge_lock to use static initializers and removed unused functions sge_setup_lock_service() etc. o fix for subscribe/unsubscribe after being registered Bugster: 6453426 Review: RD AH-2006-10-26-2: Bugfix: Fix an issue with cl_com_set_synchron_receive_timeout() be called with a NULL commlib handle upon changes with sched_conf(5) Review: AA AH-2006-10-26-1: Enhancem. Added commlib wrapper functions to allow qmaster message buffer be monitored through dtrace Review: CR AH-2006-10-26-0: Bugfix: qmaster loggings "scheduler tried to remove a incomplete" and "scheduler send a order for a changed user/project" Bugster: 6486125 Issue: 2108 Review: JG AA-2006-10-25-0: Enhancem. o openss-0.9.7l o aimk fix for darwin o execd USE_QSUB_GID problem fix o drmaa build junit.jar dependency o darwin JNI_CreateJavaVM_Impl due to missing JNI_CreateJavaVM o jni.h wrong dependency Review: JG RD-2006-10-24-0: Enhancem. o added lirs qconf options to qconf(1) o added qlimit(1) manpage o added sge_limitation_rule_set(5) man page o unified format of all man pages Review: DT AH-2006-10-23-1: Bugfix: Immediate array jobs that exceed the number of available resources are not rejected completely Bugtraq: 6480138 Review: AA AH-2006-10-23-0: Bugfix: Discrepancy between load values reported by Gridengine and from the HP-UX 64 bit env Bugtraq: 6470048 Review: JG RH-2006-10-23-0: Bugfix: jgdi: Cat and paste error prevents handling of ExecHostAddEvent Review: AA GH-2006-10-20-0: Bugfix: Account string does not accept the "|" character Bugtraq: 6475282 Review: AH RD-2006-10-19-0: Bugfix: - Execd ignores in certain cases JOB_EXECUTION request - removed some redundant code Bugster: 6483941 Issue: 2106 Review: JG AH-2006-10-19-0: Bugfix: Ensure events emitted upon qmod -d/-e get always sent as a whole to the scheduler. Review: RD JG-2006-10-19-0: Bugfix: Fixed broken build on irix and darwin. Review: RD AH-2006-10-16-3: Enhancem.: Added RUNPATH link flag to SHARED_LFLAGS to fix problem of libspoolb.so depending on LD_LIBRARY_PATH for libdb-4.4.so. Bugster: 6456290 Issue: 2090 AH-2006-10-16-1: Enhancem.: Fixed NO_SGE_COMPILE_DEBUG setting to cause rmon-macros be inactive in pack.c, cull_list.c and cull_hash.c AH-2006-10-16-0: Enhancem.: Forgot inst_common.sh with AH-2006-10-12-0 Bugster: 6456290 Issue: 2090 Review: JG AH-2006-10-12-0: Enhancem.: Added distinst -shlibpath to allow enforcement of LD_LIBRARY_PATH setting even with Solaris/Linux where actually $ORIGIN relative RUNPATH is used instead Bugster: 6456290 Issue: 2090 Review: AS AH-2006-10-10-0: Enhancem: Added #pragma no_inline() to prevent inlining of some functions as required for dtrace(1) pid provider. This change is explained in dtrace-doc CR #6480235. Review: RD RD-2006-10-09-0: Bugfix: Fixed build errors - with non gnu make - on darwin_x86 - make rules for windows Review: AA JG-2006-10-09-0: Bugfix: Fixed msg_juti.h. aimk didn't fail on windows in case of errors while building windows applications. Review: RH AA-2006-10-09-0: Enhancem: o corrections in LD_LIBRARY_PATH in run_util.sh for jobmon Review: RH HP-2006-10-06-0: Enhancem: o Windows port of authuser Review: TBD by RH AA-2006-10-06-0: Enhancem: o reordering of arguments, formatting o build.xml test debug fix o rollback of NEW_GDI_STATE in aimk Review: RD RH-2006-10-06-0: Enhancem: juti: GridCA.renewCaCertificate implemented Review: MD RD-2006-10-05-0: Bugfix: fixed locking issue introduced by RD-2006-10-04-0 Review: JG RD-2006-10-04-0: Bugfix: o Added reference check for consumable in limitaition rule set o Added reference check for consumable in load_fomula o Optimized qmaster locking Bugtraq: 6477176 Issue: 2104 Review: JG RH-2006-09-26-0: Enhancem: o some missing events for jgdi implemented o rmon_condition does not use rmon context Review AA AA-2006-10-04-0: Enhancem: o gdi_state part of context o aimk additional java target fixes, NEW_GDI_STATE define Review: RD DT-2006-10-02-0: Docs: o Added -DDRMAA_10 info to man pages Bugtraq: 6474739 AA-2006-10-02-0: Enhancem: o jgdi qlimit o sge_error_verror() o japi segv o copyright comments Review: RH RD-2006-09-29-0: Enhancem: o category string considers limitation rule sets for -u switch Cleanup: o optimized use of dstring for performance Review: AA AA-2006-09-28-0: Cleanup: o prototype cleanup o correction of Makefile o removal of obsolete files Review: RD AH-2006-09-27-0: Enhancem: Adjusted snapshot.csh so that experimental and testuite directories are no longer in src tar-balls Review: JG AA-2006-09-27-0: Bugfix: o qstat ok output o sge_get_master(int-> bool) o async gdi, gdi_state o qlimit Makefile rule Review: JG AH-2006-09-26-1: Bugfix: Disabled RMON_USE_CTX in rmon_condition() RD-2006-09-26-0: Bugfix: fixed java doc install Review: AA AH-2006-09-26-0: Bugfix: Do bootstrap_get_default_domain() in sge_hostcpy() if required only AA-2006-09-26-0: Enhancem: o removed obsolete qstat_filter.[ch] and build rule in Makefile o missing error message for non admin/submit hosts in qstat o error handling in qacct o move of clients/qlimit/qlimit_output.* to clients/common/sge_qlimit.* for jgdi Review: RD RD-2006-09-25-0: Bugfix: o added SGE_GDI_REPLACE o fixed lirs scheduling issue o corrected some qconf return values o wrong gdi command killed qmaster o fixed single quote issue Bugster: 6342224, 6349614, 6403155 Issue: 1850, 1889, 2101 Review: AA RC-2006-09-22-0: WWW o fixed typo in sge_migrate.html AA-2006-09-21-0: Bugfix: o wrong prototype fixed in centry_redebit_consumables() o error handling in unpacking code o formatting Review: RD RH-2006-09-20-1: Bugfix: o sge_ca does not longer use 'global' tmp files o GridCAImpl writes lock file before calling sge_ca Review: HP JG-2006-09-20-0: Bugfix: Fixed build on tru64. Cleanup. Testsuite: Fixed aimk clean call (falsely reported errors). Made project_access_list test available for 6.x. Review: RD RH-2006-09-20-0: Enhancem: o new script scripts/compilearch with determines BUILDARCH, COMPILE_ARCH and TARGET_BITS Review: AA AA-2006-09-18-1: Enhancem: o bugfix, profiling was switched off Review: RH RH-2006-09-18-0: Enhancem: o sge_ca checks whether needed binaries are installed Review: AA JG-2006-09-19-0: Enhancem.: SPARC Linux port. Testsuite: Fixed writing to mailfile instead of sending mail. Review: RD DT-2006-09-18-0: WWW o Backed out DT-2006-09-16-0 AA-2006-09-18-0: Enhancem: o cleanup of build process of drmaa,juti,jgdi o build source/target for cullconv o bug in sge_centry_del() that leads to master abort for qconf -mc in delete case Review: RH DT-2006-09-16-0: WWW o Added Google Analytics tracking code to html AA-2006-09-15-0: Enhancem: o introduced sge_gdi_ctx_class_t for jgdi support sge_env, prog_state, path_state, bootstrap_state, sge_csp, commlib handle, cert info to make this opaque and get rid of global variable dependencies o the code is currently activated by -DTEST_GDI2 -DTEST_QMASTER_GDI2 o qstat, qhost restructuring, separation of algorithm and output functionality o java part of jgdi o cvs tag to checkout latest maintrunc without changes: GDI2_TAG JG-2006-09-12-0: Bugfix: Intermediate accounting records for pe slave tasks contained invalid submission time. Bugtraq: 6457900 Issue: 2092 Review: RH SG-2006-09-05-0: Enhancm.: added a document describing the profiling output JG-2006-09-01-0: Testsuite: Fixed a TCL error in SGE 5.3 related code. Fixed broken ADOC headers. JG-2006-08-31-1: Testsuite: Always add a build number to builds done by testsuite. Minor fix for setups without root access. Review: CR JG-2006-08-31-0: Enhancem.: Added means to output a build number with the version string. Cleanup: Fixed broken ADOC headers. Review: SG AH-2006-08-30-1: Bugfix: Fix problem with classic spooling write object return codes not be checked Review: JG AH-2006-08-30-0: Bugfix: - Fix problems with qsub -w e|w|v resulting from uninitialized hgrp_list and lirs_list in sge_assignment_t - Fix issue with cl_com_ssl_func__ERR_get_error being NULL RH-2006-08-18-2: Cleanup: o invalid pathes in aimk.site RH-2006-08-18-1: Cleanup: o wrong checkin (sge_passwd.c) RH-2006-08-18-0: Enhancem: o i18n for juti o strong argument checking for GridCAImpl.initCA o aimk.site contains JUNIT_JAR parameter Review: MD RD-2006-08-16-0: Enhancem: - rewrite of limitation rule matching code to handle all kind of wildcard searches - added module test for limitation rule matching code - added module test for resource utilization - added generic functions for qlimit/qsub/qstat request files Changed: qlimit, qstat, qsub, qmaster, schedd Review: AA HP-2006-08-11-1: Enhancem: Improved starting and stopping of N1 Grid Engine Helper Service during test. Changed: checktree/system_tests/windows/gui/check.60.exp HP-2006-08-11-0: Bugfix: Now loading key and rand file for Administrator, not longer for HOST+Administrator. Changed: libs/gdi/sge_security.c Review: JG RD-2006-08-11-0: Enhancem: Added reservation code for limitation rule sets Moved gdi version defines to libs/gdi/version.c Changed: all Review: SG JG-2006-08-11-0: Testsuite: Minor fixes and cleanup. Review: RD MD-2006-08-10-0: Bugfix: The autoinstall need ssh daemon when the autouninstall is done local Issue: 2095 Bugtraq: 6453427 Changed: install Review: SG JG-2006-08-10-1: Bugfix: Fixed memory leaks and access errors. Review: AH, RD JG-2006-08-10-0: Testsuite: Introduced host configuration parameter "java compile host" Allow multiple code coverage methods to be used in a single testsuite run (e.g. for c and for java). Review: RD RD-2006-08-10-0: Enhancem: Raised gdi version to be incompatible with 6.0 Added 6.0u8_2 to gdi version array Review: JG AH-2006-08-08-0: Enhancem.: Prepared relative path $ORIGIN/../../lib/$buildarch be used rather than LD_LIBRARY_PATH for Solaris/Linux Bugster: 6456290 Issue: 2090 Review: AA, JG RD-2006-08-07-0: Enhancem.: Enabled category scalability improvement for usersets and projects (was disabled by RD-2006-08-03-0) Review: AH EB-2006-08-07-0: Enhancem.: Enabled pam support in juti for DARWIN_PPC AND DARWIN_X86 Review: AA JG-2006-08-04-0: Testsuite: Added function parse_properties_file to parse Java properties files, as well as the execd/shepherd config or environment file. Minor fixes and cleanup. Review: RD RD-2006-08-03-1: Enhancem.: o Added qlimit Changed: qlimit Review: AA RD-2006-08-03-0: Enhancem.: o Added limitation rule set scheduling code for - sequential jobs - parallel jobs - dynamical limits o Added qmaster limitation rule set verification o Added qmaster debiting code o fixed some minor memory leaks and bugs Changed: all Review: JG HP-2006-08-03-1: Enhancem.: Added a reserved complex variable "display_win_gui" that allows submitter to request the GUI of the job to be displayed to the visible desktop and allows the scheduler to make the right decisions. Added a test for it to the testsuite Changed: execd, shepherd, testsuite Added: centry/display_win_gui, system_tests/windows/gui test Bugster: 6428621 Review: JG, MD HP-2006-08-03-0: Bugfix: o fixed mixed short/long hostnames for ignore_fqdn=false systems in load report Changed: interix-loadsensor.sh Review: JG CR-2006-08-03-0: Bugfix: o fixed bug in sge_ca script Review: RH RH-2006-08-02-0: Enhancem.: o multipurpose sge_ca script o bugfixes in juti Review: AA AH-2006-08-02-0: Enhancem.: Added comment to dist/util/arch about alternative means for determining GNU C library version. AA-2006-08-01-0: Enhancem.: Preliminary version of LIRS code for qmon added Review: RD RD-2006-07-31-2: Cleanup: Fixed gcc42 compiler warnings Review: AA RD-2006-07-31-1: Bufix: Fixed broken build with option -spool-classic Review: JG RD-2006-07-31-0: Testsuite: Added basic functions for limitation rule sets fixed some testsuite tests Review: JG JG-2006-07-31-0: Testsuite: Minor fix for qrsh check (qconf vi handling) Review: RD HP-2006-07-26-0: Bugfix: Fixed broken build due to adding of juti/removing of classes Changed: scripts/zerodepend, aimk, Makefile.global, juti RH-2006-07-26-0: Enhancem.: First version of libs/juti added Review: AA JG-2006-07-25-0: Bugfix: sge_share_mon did not work with automatic policy enforcement ("default" nodes in sharetree). Bugfix: sharetree node names in sge_share_mon and in reporting did not contain the whole path. Merge of JG-2006-07-24-0 from V60s2_BRANCH. Bugfix: sge_share_mon was broken (missing MT init). Bugster: 6448704, 6449439 Issue: 2167 Review: RD JG-2006-07-25-0: Testsuite: Use sharetree infrastructure provided by JG-2006-07-24-0 in bugs/issuezilla/1848. Extended qconf test to test creation, modification, and deletion of sharetree, as well as adding projects. Add generic error messages (user/host permissions) to handle_sge_errors. Review: RD JG-2006-07-24-0: Testsuite: Added test for sge_share_mon. Added utility functions to create and modify sharetree. Use these functions in various checks instead of calling and handling qconf. Fixed get_ps_info for darwin. Fixed qmod_general test. Have testsuite unique tmpdir in queue config to avoid problems with multiple testsuites running in parallel on the same hosts. Review: RD AH-2006-07-19-0: Testsuite: Enhanced system_tests/scheduler/queue_requests to cover -q cqueue_list -masterq cqueue_list -soft -q qinstance_list -soft -q cqueue_list IZ: 2085 Review: JG RD-2006-07-17-0: Bugfix: - Fixed berkeleydb spooling and CSP mode on freebsd - removed libssl.bundle for MacOS (no longer necessary since MacOS 10.4) Review: AA JG-2006-07-11-0: Bugfix: Fixed typos in complexes man page. Fixed version mismatch in jni build on darwin. Review: RD AA-2006-07-05-0: scripts: Added OPENSSLSOVERSION to distinst.site to support use of other SSL versions Review: RD Changed: scripts/distinst[.site] aimk.site AH-2006-07-04-0: Bugfix: Fixed problem with host_update_categories() being wrongly called upon qconf -dh/-ds due to AH-2006-06-23-1 CR-2006-07-04-0: Testsuite: fixed bug when only 1 execd is set in configuration fixed bug when adding (discovering) a new host into testsuite host configuration file CR-2006-07-03-0: Testsuite: new menu item 50 - compile and install only checktree hooks Review: JG AA-2006-06-30-0: Bugifx: qping segfaults in ssl mode CR: 6445469 IZ: 2081 Review: CR CR-2006-06-30-0: Testsuite: added java15 settings to host configuration added passwd hook for additional checktree Review: JG JG-2006-06-28-0: Testsuite: Add execd_param required for windows domain users in install_core_system check. Fixes and cleanup for qconf and qrsh check. Cleanup of testsuite password query. Review: CR AA-2006-06-28-0: Testsuite: test_category fixed Review: AH Changed: test_category OJ-2006-06-27-0: Testsuite: Added test case for qstat -ext, qstat -f -ext, qstat -t, qstat -pri, qstat -urg, qstat -j ERROR, qstat -f -r, qstat -f -urg, qstat -f -qs E, qstat -j. with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT AA-2006-06-27-0: Cleanup: Removal of unused zlib and XtReplay 3rdparty code Review: AH Changed: qmon JG-2006-06-26-0: Testsuite: Added support for windows hosts. Review: CR AH-2006-06-23-1: Docs: Fixed sge_conf(5) reprioritize and related paragraph in queue_conf(5) Review: AA Changed: man pages AH-2006-06-23-1: Enhancem. Merged scalability enhancements from S3 branch AH-2006-06-23-0: Enhancem. Forgot to check-in aimk changes required for AH-2006-05-29-1 JG-2006-06-23-0: Enhancem.: Build on darwin-x86 platform (fixed duplicate symbols in spooling libs). Build on tru64. Review: RD AH-2006-06-22-1: Enhancem.: Completized qping output for job execution requests sent from qmaster to execd Review: CR JG-2006-06-21-0: Enhancem.: Separate darwin platform into darwin-ppc and darwin-x86. darwin-x86 port still to be done! Fixed build for MacOS 1.4 (ppc) with gcc-4. Review: RD DT-2006-06-20-0: Testsuite: o Fixed test 1529 o Improved test_drmaa Changed: test_drmaa, test_drmaa_no_bin OJ-2006-06-15-0: Testsuite: Added test case for qstat -explain E, qstat -ne, qstat -f -q all.q, qstat -f -ne, qstat -l q=all.q, qstat -s r, qstat -s s, qstat -s z, qstat -s hs, qstat -s hu, qstat -u wrong_user, qstat -U $CHECK_USER with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT RD-2006-06-14-0: Bugfix: fixed several potential security issues Bugtraq: 6438475 Changed: cull Review: JG,AA RD-2006-06-13-0: Enhancem.: - new qconf switches for limitation rule sets - new GDI calls for limitation rule sets Changed: aimk, qmaster, qconf Review: JG HP-2006-06-08-1: Enhancem.: Build targets depend and clean are now properly for- warded to Makefiles for Win32 applications. Changed: aimk HP-2006-06-08-0: Enhancem.: Added man page for SGE_Helper_Service.exe OJ-2006-06-06-0: Testsuite: Added test case for qstat -g c, qstat -g d, qstat -g t, qstat -s a, qstat -s p, qstat -u user with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT HP-2006-06-02-0: Bugfix: Fixed wrong function declaration Changed: sge_uidgid.c JG-2006-06-01-0: Testsuite: Minor enhancements, fixes, cleanup. Review: RH AH-2006-05-31-0: Enhancem.: analyze.rb utility can now deal with array jobs and rescheduled jobs. AH-2006-05-29-2: Cleanup: Removed pointless code in qdel Review: JG AH-2006-05-29-1: Bugfix: Enhanced build process to set library name (soname) with 0.95/1.0 libdrmaa for Solaris and Linux. Changed: libdrmaa Issue: 2059 Bugtraq: 6429305 Review: JG JG-2006-05-26-0: Testsuite: Made testsuite host configuration Grid Engine version aware. It is now possible to use one host config file for various Grid Engine versions. The architecture string and the compile host property for a host are Grid Engine version dependent. Cleanup and minor fixes. Review: CR AH-2006-05-23-0: Bugfix: Changed drmaa.c to ensure there are no DRMAA 1.0 specific linking symbol in libdrmaa.so.0.95. As a result DRMAA applications build with 1.0 will not start any longer if LD_LIBRARY_PATH points to 0.95 library. Review: RD Changed: libdrmaa Issue: 2059 Bugtraq: 6429305 HP-2006-05-23-0: Cleanup: Removed unneccessary dependencies from libdrmaa, removed libjapi.a, removed unneccessary dependencies of libjapi.a from several binaries. Review: AH Changed libdrmaa Makefile, qsub Makefile AH-2006-05-20-0: Bugfix: Added shared library dependency to libdrmaa.so build rule to overcome problem of -ldl -lpthread required at stage for DRMAA applications. Review: AA, JG Changed: libdrmaa Issue: 2057 Bugtraq: 6428050 JG-2006-05-19-0: Testsuite: Prepared testsuite framework for N1GE 6.5. Enhancements and cleanup of qsub test. Review: CR HP-2006-05-18-0: Enhancem.: Added support for SGE_Helper_Service.exe on Windows Changed: sge_execd, sge_shepherd, libwingrid, libuti, distinst Bugtraq: 6363237 Review: MD, CR JG-2006-05-17-0: Bugfix: Jobs with negative -p priority were rejected by sge_qmaster. Issue: 2055 Bugtraq: 6424565 Fixed a memory leak in execd job request verification. Enhancem.: Jobs with positive -p priority submitted by a non operator user are rejected. In earlier versions, the priority is silently set to 0. Review: RD RD-2006-05-17-0: Enhancem.: - Added flatfile and classic spooling code for Resource Limitation Rules Bugfix: - qconf rejects configuration, when attribute value ends with a space character - fixed two cases of reading past allocated memory Bugtraq: 6407525 Issue: 2023 Review: JG JG-2006-05-15-0: Enhancem.: Improved verification of incoming job requests in execd. Review: RD OJ-2006-05-12-0: Testsuite: Added test case for qstat -F, qstat -F rerun,h_vmem with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT JG-2006-05-10-0: Testsuite: Fixed incorrect parsing of qstat output. Issue: 2052 Review: RD MD-2006-05-08-0: Bugfix: during QMaster installation, creation of local database directory fails on hp11 Issue: 2051 Bugtraq: 6412865 Changed: install Review: HP JG-2006-05-02-0: Enhancem.: Use Berkeley DB 4.4.20 for berkeleydb spooling. Port to Mac OS 10.4, enable build with gcc-4. Review: RD OJ-2006-04-26-0: Testsuite: Added test case for qstat -f with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT JG-2006-04-26-0: Testsuite: Minor enhancements, fixes, and cleanup. Review: AA RD-2006-04-19-0: Bugfix: fixed qmon for lx24-ia64 disabled optimization because compiler is broken Changed: all Review: AA, JG JG-2006-04-19-0: Testsuite: Minor fixes and cleanup. Review: RD CR-2006-04-19-0: Cleanup: cleanup of RAND file reading in commlib SSL mode Changed: cl_ssl_framework.c => commlib (merge from V60s2_BRANCH) Review: AA CR-2006-04-18-0: Bugfix: added fix for aix and hp in ssl module Changed: cl_ssl_framework.c => commlib (merge from V60s2_BRANCH) Review: RD RH-2006-04-12-0: Bugfix: The man page QUEUE_CONF(5) does not describe the memory specifier 'G' Issue: 2029 Bugtraq: 6411660 Review: RD AH-2006-04-11-1: Bugfix: Job sequence number got screwed up when restarting qmaster daemon Issue: 2028 Bugtraq: 6411230 Review: RD,JG JG-2006-04-11-0: Testsuite: Added test for issue 2028. Review: RD MD-2006-04-07-0: Testsuite: Issue_1330 test. Check for existing bdb spooldir Review: JG JG-2006-04-07-0: Testsuite: Minor fixes and cleanup. Review: RD RD-2006-04-06-0: Bugfix: execd on windows can't to connect to qmaster (introduced by security enhancm. RD-2006-03-17-0) Review: MD AH-2006-04-06-1: Bugfix: Fix reservation for array tasks (CR 6260656) broken by SG-2006-03-29-0 Review: RD,JG AH-2006-04-05-0: Bugfix: Fix schedd abort() upon qconf -ks Review: RD RD-2006-04-04-0: Bugfix: CSP installation with admin user = root broken fixed broken build on AIX43 changed version to 60u8 in scripts Issue: 2026 Bugtraq: 6408109 Review: CR JG-2006-04-04-0: Bugfix: Fixed too restrictive job name verification. Testsuite: Fixed broken testsuite. Review: RD OJ-2006-04-03-0: Testsuite: Added test case for qstat with XML parsing. in client/qstat. Added new file tcl_files/parser_xml.tcl with new XML parsing procedures. Issue: 1206 Review: DT JG-2006-04-03-0: Bugfix: Scheduler tuning during qmaster installation broken. Testsuite: Added testsuite test for IZ 2022. Changed default testsuite product_type to "sgeee". Issue: 2022 Bugtraq: 6407523 Review: MD RD-2006-04-03-0: Testsuite: Enabled uncommented tests Review: JG SG-2006-04-03-0: Bugfix: Scheduler hangs after a qmaster crash and restart Issue: 2021 Bugtraq: 6407513 Review: RD JG-2006-03-31-0: Bugfix: fixed scheduler memory corruption (core dump) introduced by MD-2006-03-28-0. Review: RD SG-2006-03-29-0: Bugfix: fixed the qstat.xsd schema Issue: 2017 Bugtraq: 6405794 Review: RD RD-2006-03-29-0: Enhancem.: Improved verification of incoming job requests in qmaster (JG-2006-03-17-0 continued) Review: AA SG-2006-03-29-0: Bugfix: - Unclear share caclulation of running jobs - Job tickets are not correctly shown in qstat for none running jobs - slow scheduler performance for jobs with hard queue requests - Tickets are not reseted for running jobs after disableing the ticket policy Issue: 1550, 1956, 1981, 2003 Bugtraq: 6291033, 6368747, 6384709, 6398723 Changed: scheduling Review: RD AA-2006-03-28-0: Enhancem.: - feature_activate code removed - handling of sge_get_auth_info - sge_remove_event_client() index out of bounds - EV_update_func check - empty_job_list_filter check Review: RD MD-2006-03-28-0: Bugfix: schedulers mem use growing, if pe jobs are running Issue: 1980 Bugtraq: 6384698 Changed: scheduling Review: HP MD-2006-03-27-0: Bugfix: RPC Berkeley DB install failed due to FQDN hostname Issue: 1985 Bugtraq: 6380207 Changed: install Review: HP JG-2006-03-27-0: Enhancem.: Improved verification of incoming event client requests in sge_qmaster (merged JG-2006-03-24-0 from V60s2_BRANCH). Review: AA MD-2006-03-24-0: Bugfix: fixed broken build MD-2006-03-23-0: Enhancem.: - security enhancement, improved string verification. Review: RD RD-2006-03-22-1: Bugfix: - qrsh does not accept -o/-e/-j - qrsh issue with interactive jobs and directory write permissions - interactive jobs leave behind output/error files if prolog/epilog are run Issue: 1997, 1998, 368 Bugtraq: 6391238, 6390494, 4737342 Review: JG JG-2006-03-22-1: Testsuite: Insure code coverage analysis continued, cleanup and minor fixes. Review: RD RD-2006-03-22-0: Bugfix: qstat -u crashes Bugtraq: 6401993 Issue: 2010 Review: JG JG-2006-03-22-0: Enhancem.: Improved verification of incoming petask requests in sge_execd (merged JG-2006-03-21-0 from V60s2_BRANCH). Review: RD RD-2006-03-20-1: Bugfix: fixed broken build RD-2006-03-20-0: Enhancem.: Added additional permission checking in CSP mode (Merge of V60s2_BRANCH, RD-2006-03-17-0) Bugtraq: 6400729 Review: AA JG-2006-03-20-1: Bugfix: Fixed a deadlock in qmaster when writing (closing) the reporting file failed. Bugtraq: 6397383 Issue: 2007 Review: RD JG-2006-03-20-0: Enhancem.: Added verification of incoming job and petask request in qmaster and execd (no full check yet). (Merge of V60s2_BRANCH, JG-2006-03-17-0) Review: AA AA-2006-03-17-0: Bugfix: several buffer overruns, merge from V60s2_BRANCH Bugtraq: 6397987 Review: RD Changed: all SD-2006-03-16-0: Bugfix: Fixed typo compile problem in SGE PQS API Issue: 2006 AH-2006-03-16-0: Bugfix: Applied patch to fix qstat -j XML output for mail list. Issue: 2005 Review: Craffi, MD OJ-2006-03-14-0: Testsuite: Moved purge_queue in sge_queue.60.tcl, moved test cases with -purge queue in check.60.exp; fixed bugs in purge_queue (). Issue: 477 Review: DT CR-2006-03-14-0: Cleanup: Off-by-one overrun in communication library (merged from V60s2_BRANCH) Bugtraq: 6398008 Review: RD Changed: commlib CR-2006-03-13-0: Cleanup: code cleanup and logging in commlib ssl module (merged from V60s2_BRANCH) Review: AA RC-2006-03-12-0: Bugfix: make sure SGE is always 32-bit on PPC Linux Review: AH Changed: aimk JG-2006-03-10-0: Testsuite: fixes for file_parsing and complex tests. Review: RD RD-2006-03-10-0: Testsuite: fixed mod_queue with fast_add option Review: JG RC-2006-03-10-0: Enhancem.: allow execd to run on Scyld Beowulf - Contributed by RH (Scyld also provided the same fix) Issue: 1936 Changed: execd Review: RC MD-2006-03-09-0: Bugfix: The parallel automatic install may overload machines Issue: 1986 Bugtraq: 6387371 Changed: install Review: HP RD-2006-03-09-0: Bugfix: - fixed race conditions for sge_ca -renew - fixed qmaster core dump with invalid revocation list Changed: sge_ca, commlib Review: AA OJ-2006-03-06-0: Testsuite: Moved purge_queue in sge_queue.60.tcl, moved test cases with -purge queue in check.60.exp; fixed bugs in purge_queue (). Issue: 477 Review: DT MD-2006-02-22-0: Bugfix: execd installation on Windows fails due to missing spooldefaults Issue: - Bugtraq: 6376809 Changed: install Review: RD DT-2006-03-02-0: Bugfix: o Fixed errors in DT-2006-03-01-0 Changed: JAPI, testsuite DT-2006-03-01-0: Enhancem.: o Completed DRMAA 0.95 => 1.0 o Made DRMAA sessions reconnectable Bugfix: o Fixed illegal memory access o Fixed DRMAA implementation string o Added missing man page o Removed INIT_ALPP macro o Fixed DRMAA memory leak Issue: 1740, 1993, 1994 Bugtraq: 6365045, 6391930, 6391947 Changed: JAPI, man pages, testsuite, qsub, qconf Review: AH OJ-2006-02-27-0: Testsuite: Restored old calls to set_cqueue_default_values set_cqueue_specific_values, in sge_queue.60.tcl; fixed mod_queue Review: DT RC-2006-02-25-0: Bugfix: mem_free 4GB wraparound on MacOSX Changed: libs/uti/sge_loadmem.c Review: HP HP-2005-02-24-0: Bugfix: on some Windows execution hosts, execd hangs after the job has finished Bugtraq: 6363245 Changed: source/clients/qacct/qacct.c source/common/basis_types.h source/daemons/execd/reaper_execd.c source/libs/gdi/sge_gdi_request.c source/libs/uti/sge_uidgid.c source/libs/uti/sge_uidgid.h Review: CR RD-2006-02-24-0: Testsuite: fixed several testuite tests Review: JG RD-2006-02-23-0: Bugfix: Merged EB-2005-12-15-0 from V60s2_BRANCH subordinate properties lost on qmaster restart Issue: 1741 Bugtraq: 6319223 Changed: sge_qmaster Review: SG RD-2006-02-23-0: Bugfix: commlib closes wrong connection on SSL error performance enhancement for certificate revocation Issue: 1990 Bugtraq: 6389526, 6387206 Review: CR RD-2006-02-23-0: Testsuite: Fixed broken install_core_system Review: CR MD-2006-02-22-0: Bugfix: - "qstat -j" aborts Issue: 1972 Bugtraq: 6384682 Changed: qstat Review: JG HP-2006-02-22-0: Bugfix: Fixed broken update_ts_config_version Removed unused string from msg_daemons_common.h Changed: testsuite/check.exp daemons/common/msg_daemons_common.h Review: CR JG-2006-02-22-0: Testsuite: New check system_tests/scheduler/queue_requests, covering also IZ 1981. Review: SG HP-2006-02-21-0: Bugfix: Fixed broken build Changed: libs/japi/Makefile DT-2006-02-17-0: Bugfix: o Added missing howto info. - Contributed by Chris Dagdigian Changed: www Review: DT OJ-2006-02-16-0: Testsuite: Added test cases for *prj, *q, *u, *sconf for qconf in sge_queue.60.tcl, sge_host.tcl, sge_sched_conf.tcl sge_users.tcl, sge_project.tcl as error handling for these tests. Review: DT JG-2006-02-16-0: Enhancem.: Enhanced the DRMAA example "compile" to delete the compile jobs at CTRL-C. Review: CR RD-2006-02-14-0: Bugfix: Interrupting qrsh while pending does not remove job Bugtraq: 6287945 Issue: 1639 Review: CR RD-2006-02-13-1: Bugfix: qstat produces non-well-formed XML output Bugtraq: 6384812 Issue: 1957 Review: JG RD-2006-02-13-0: Bugfix: resource filtering in qselect broken Bugtraq: 6383513 Issue: 1977 Review: JG OJ-2006-02-10-0: Testsuite: Added test cases for *hgrp, *prj for qconf in sge_host.60.tcl, sge_project.tcl as well as error handling for these tests. Review: DT JG-2006-02-10-0: Testsuite: Minor fixes and cleanup. Review: RD HP-2006-02-09-0: Bugfix: qrsh now adds a '\n' to error output recieved from shepherd. While fixing this, I replaced some strcpy()/strcat() with snprintf(). Bugtraq: 6383256 Issue: 1978 Changed: qsh.c, err_trace.c, qlogin_starter.c Review: RD OJ-2006-02-06-0: Testsuite: Added error handling in *ckpt, *e routines in sge_checkpoint.60.tcl, sge_host.tcl and control_procedures.tcl Review: DT JG-2006-02-03-0: Testsuite: Improved error handling for sge functions called from testsuite. Code coverage analysis using Solaris cc/tcov. Fixed testsuite option execute_func and qping test. Review: RD OJ-2006-01-31-0: Testsuite: Added error handling in *attr routines in sge_procedures.tcl; modified *cal routines in sge_calendar.tcl Review: DT Issue: 477 JG-2006-01-31-0: Enhancem.: Merged JG-2006-01-30-0 from V60s2_BRANCH: cleanup aimk code coverage options cleanup of daemonizing including insure related actions. Review: RD AH-2006-01-26-0: Enhancem.: Added more relevant information to starting up info message in daemon messages files Changed: sge_qmaster, sge_execd, sge_schedd Review: JG JG-2006-01-25-0: Testsuite: Fixes and cleanup in testsuite framework. Review: RD OJ-2006-01-18-0: Testsuite: Added -M(m)ckpt functionality to in sge_checkpoint.tcl, modified set_exechost.tcl to support fast add Review: DT, JG Issue: 477 JG-2006-01-16-0: Testsuite: Enhanced testsuite framework by "busy connection" detection (missing close_spawn_process calls). Additional menu item dumping open connections. Cleanup in remote_procedures. Review: CR OJ-2006-01-13-0: Testsuite: Fixed return codes for *attr procedures in sge_procedures.tcl Review: pending Issue: 477 JG-2006-01-12-0: Testsuite: Fixed broken testsuite (install). SG-2006-01-12-0: Bugfix: fixed SG-2006-01-06-0, local event clients did not work Changed: qmaster Review: JG JG-2006-01-11-0: Testsuite: Cleanup of remote_procedures. Review: CR OJ-2006-01-11-0: Testsuite: cleaned up add_* del_* procedures for qconf tests Changed: testsuite Review: JG Issue: 477 SG-2006-01-06-0: Enhancem.: enables process loal event clients enables 1 event client per thread Changed: all Review: JG DT-2006-01-04-0: Docs: o Updated qselect man page Issue: 1955 Bugtraq: 6368942 Changed: qselect(1) DT-2006-01-04-0: Testsuite: o Enhanced drmaa_wait() tests Changed: test_drmaa SG-2006-01-04-0: Bugfix: fixed the build OJ-2006-01-03-0: Testsuite: added stnode, event client functionality to qconf test added sge_sharetree*, sge_event_client in tcl_files Changed: testsuite Review: pending Issue: 477 JG-2005-12-28-0: Bugfix: - qrsh always did verbose output (no CR/IZ, as the bug did not appear in any released product). Testsuite: - fixed broken testsuite for 5.3 - shadow only hosts were not added to admin host list - enhancements to remote connection handling Review: CR DT-2005-12-23-0: Enhancem.: o Cleaned up DRMAA 0.95/1.0 transition o Moved ARCo utils into libutil o distinst defaults to current architecture Review: pending (AH) Changed: JAPI, classes, distinst, aimk JG-2005-12-23-1: Testsuite: Fixed broken testsuite. RD-2005-12-23-0: Bugfix: modification of hgrp results in glibc error message printed by qconf and abort Bugtraq: 6364440 Issue: 1947 Review: JG JG-2005-12-23-0: Bugfix: - fixed arch script for ia64 architecture - fixed a minor memory access problem in japi - fixed a bug in rsh client Bugtraq: 6366691 Testsuite: make testsuite run on lx24_ia64 and lx26_ia64. Review: RD OJ-2005-12-22-0: Testsuite: moved procedues in sge_pe, sge_host, sge_calendar, sge_qconf_config, sge_users, sge_userset Review: DT Issue: 477 OJ-2005-12-22-1: Testsuite: added files sge_calendar.*, sge_users.* sge_userset.* which contain procedures to be used by new qconf tests. Review: DT Issue: 477 OJ-2005-12-22-2: Testsuite: new procedures to test qconf -pe Review: DT Issue: 477 OJ-2005-12-22-3: Testsuite: new files from sge_procedures.* containing procedures for checkpointing. Review: DT Issue: 477 OJ-2005-12-22-4: Testsuite: new procedure for qconf -sconfl Review: DT Issue: 477 OJ-2005-12-22-4: Testsuite: new procedures for qconf Review: DT Issue: 477 OJ-2005-12-22-5: Testsuite: new qconf tests in clients/qconf Review: DT Issue: 477 OJ-2005-12-22-6: Testsuite: new procedure for qconf -sprjl Review: DT Issue: 477 JG-2005-12-21-0: Bugfix: - fixed broken build - fixed broken distinst for multiple architectures Enhancem.: - compilation with code coverage (tcov) on Solaris Testsuite: - don't create error/output files in qacct test Review: SG DT-2005-12-20-0: Bugfix: o Fixed broken build Changed: japi/Makefile RD-2005-12-20-0: Bugfix: qsub -w w changes -sync behavior Issue: 1945 Bugtraq: 6363823 Review: SG CR-2005-12-20-1: Cleanup: fixed potential buffer overflows in sge_exec_job() Issue: 1950 Bugtraq: 6365380 Review: RD CR-2005-12-20-0: Testsuite: - changed handle_vi_edit() time measurement Review: RH SG-2005-12-19-0: Enhancem.: - second part of making the libs thread safe and removing the master lists. All master lists have been removed. - fixed a bug in the event mirror. Introduced with SG-2005-12-08-0 Changed: all Review: RD DT-2005-12-17-2: Bugfix: o Fixed DENTER in sge_conf for INHERIT_ENV Changed: sge_conf.c DT-2005-12-17-1: Bugfix: o Added missing DRMAA error code Enhancem.: o Added versioning to DRMAA libs Review: pending (AH) Changed: DRMAA, distinst DT-2005-12-17-0: Testsuite: o Added missing test_drmaa tests o Added preflight check for drmaaj test o Added back excluded jUnit tests o Added suspendable sleeper script o Added env param to startup_execd{} o Added function to get shared lib var name o Added function to wait for a config change to propogate o Fixed job_environment test Issue: 1388, 1795 Review: CR Changed: testsuite CR-2005-12-16-1: Testsuite: removed waiting times in startup_qmaster() Review: TBD CR-2005-12-16-0: Testsuite: o made handle_vi_edit() more stable o additional cleanup Review: TBD EB-2005-12-16-0: Testsuite: added test for issue 1741 Review: TBD JG-2005-12-15-0: Testsuite: Extended system_tests/clients/qacct, cleanup of get_qacct function. Review: RD CR-2005-12-15-2: Testsuite: some additional code cleanup for remote procedures Review: RD CR-2005-12-15-1: Testsuite: modified issue 126 test to allow a time range for job to get into suspend state Review: RH RD-2005-12-15-0: Bugfix: qtcsh behaves differently in direct mode from qrshmode Bugtraq: 6344960 Issue: 1858 Review: CR CR-2005-12-15-0: Testsuite: - fixed some problems with handle_vi_edit() - improved shell spawn procedures - new procedure: delete_remote_file() - fixed file_parsing test - fixed qtcsh test - fixed qconf test Review: RD RD-2005-12-13-0: Bugfix: - incorrect qsub error message, if an invalid integer value is passed to the -l option - qconf -sstnode cannot find root - qstat -qs doesn't work - "qstat -explain E" displays explanation of the same error two times. - qstat -q filter does compare hosts in queue instances Bugtraq: 6329378, 6291047, 5086007, 6349037, 6362523 Issue: 1930, 751, 1941, 1933, 1940 Review: MD CR-2005-12-13-0: Testsuite: o fixed check some descriptions and added check description check at testsuite startup o fixed drmaaj test (export LD_LIBRARYN32_PATH) o install_core_system (interactive) can be aborted when running in testsuite debug mode o fixed qrsh test which sometimes killed qrsh job while qrsh_exit_status file is not written (with CTRL-C) Review: MD MD-2005-12-12-0: Bugfix: INSTALL: screen not cleared Bugtraq: 5063318 Issue: 1064 Changed: sge_ca Review: AA MD-2005-12-12-0: Bugfix: merged from S2_BRANCH incuded in u7 Changed: inst_sge, install_modules SG-2005-12-12-0: Bugfix: fixed a bug added in SG-2005-12-08-0 In case of a qconf -ke all -ks with a not existing scheduler, one would only see the scheduler error message, but not the execd messages. Changed: qmaster Review: EB HP-2005-12-09-0: Bugfix: fixed broken build on alinux Changed: daemons/qmaster/sge_qmaster_threads.c Review: Not needed CR-2005-12-09-0: Testsuite: Using longer timeouts for spool directory removal Review: RH SG-2005-12-08-0: Enhancem.: - first part of making the libs thread safe and removing the master lists - fixed a memory leak in c_gdi, related to qconf -ks - fixed a bug in job number spooling Review: DT Changed: all JG-2005-12-08-0: Testsuite: Added test cases for qconf -se and -sel. Additional utility functions and cleanup. Review: CR CR-2005-12-08-0: Bugfix: Fixed minor memory leaks from insure test Review: AA RD-2005-12-08-0: Bugfix: fixed broken shepherd on irix introduced due to FCLOSE cleanup Changed: qrsh_starter, shepherd, execd Review: CR JG-2005-12-07-0: Testsuite: New module tcl_files/sge_host.tcl. Cleanup of get_exechost. New functions for calling SGE commands and error handling. Review: CR JG-2005-12-06-0: Bugfix: parsing error in cluster config, qmaster_params Review: SG DT-2005-12-06-0: Bugfix: o Fixed DRMAA C binding version number o Fixed DRMAA memory leak Changed: DRMAA, test_drmaa Issue: 1931 Bugtraq: 6359575 Review: RD, AH RD-2005-12-06-0: Bugfix: Fixed broken bins on irix65 removed 3rdparty/snprintf from irix65 build Changed: all Review: RH HP-2005-12-06-1: Bugfix: Fixed broken build on win32-x86 Changed utilbin/Makefile HP-2005-12-06-0: Bugfix: Removed weird line of code Changed: daemons/execd/exec_job.c Bugtraq: 6291049 Issue: 1032 Review: Not needed RD-2005-12-05-0: Enhancem: returns errno message if qrsh_exit_code can't be opened added SGE_CLOSE macro Review: AA CR-2005-12-02-0: Testsuite: enhanced executable file check procedure at testsuite startup Review: AA JG-2005-12-02-0: Testsuite: Minor fixes for bugs/issuezilla/540, system_tests/scheduler/sharetree, system_tests/shadowd/migrate, auto install with bdb server, auto_install with Solaris zones. Use fast option (e.g. -Aq instead of -aq) where possible. Review: CR CR-2005-12-02-0: Testsuite: o redo of some changes by CR-2005-12-01-0 CR-2005-12-01-0: Testsuite: o Fixed some testsuite errors Review: MD DT-2005-11-30-0: Bugfix: o Fixed broken INHERIT_ENV Changed: sge_conf.c Review: pending EB-2005-11-30-0: Testsuite: create a test for issue 1877 Issue: 1879 Review: CR CR-2005-11-30-0: Testsuite: modified system_test/execd/startup test to produce warning message and skipp second startup execd test on diffent port. Test has do be re-activated when BT: 6350264 (IZ: 1893) is fixed Review: JG EB-2005-11-29-0: Bugfix: Merged following from V60s2_BRANCH: - job finish although terminate method is still running (EB-2005-11-16-1) - job finish does not terminate all processes of a job (EB-2005-11-16-0) - qconf -mstree aborts (EB-2005-11-18-0) Issue: 1874, 1876, 1877 Bugtraq: 6348299, 6348516, 6348517 Changed: qconf, sge_shepherd, sge_execd, sge_qmaster, worker Review: JG DT-2005-11-29-0: Bugfix: o Fixed broken build on Irix 6.5 Changed: shepherd Review: RD RD-2005-11-29-0: Bugfix: broken build for fbsd-i386 Review: DT CR-2005-11-29-1: Testsuite: added queue error check for checktree/system_tests/clients/qrsh/check.exp Review: RD RD-2005-11-29-0: Bugfix: deleting a qrsh job kills the rshd first fixed testsuite test qconf, file_parsing & complex Changed: shepherd Review: CR CR-2005-11-29-0: Bugfix: o check file descriptor errors after select() errors in commlib AND after get_sock_opt() errors o additional enhancment fix to CR-2005-11-16-0 Review: RD Changed: commlib JG-2005-11-28-1: Testsuite: Fixed check system_tests/shadowd/migrate. Review: CR HP-2005-11-28-0: Bugfix: removed spool* binaries and libraries from distinst Changed: scripts/distins Review: AS JG-2005-11-28-0: Testsuite: Fixed installation on lx26-* and hp11-64. Cleanup and fixes for bdb server installation. Review: CR CR-2005-11-27-2: Testsuite: modified system_tests/clients/qtcsh test to be more error tolerant Review: RD CR-2005-11-27-1: Testsuite: fixed shutdown_system_daemon() procedure for linux kernel 2.4 systems Review: RD CR-2005-11-27-0: Testsuite: fixed checktree/system_tests/clients/qconf: o removed L10N check category (test expects english qconf output) o output parsing did not work for long hostnames Review: RD CR-2005-11-26-0: Testsuite: fixed bug in issue test bugs/issuezilla/1802 Review: RD CR-2005-11-25-3: Testsuite: fixed bug in issue test bugs/issuezilla/1451 Review: RD CR-2005-11-25-2: Testsuite: qacct test depends on qsub test SG-2005-11-25-0: Testsuite: added test for issue 1916 Review: JG Changed: functional/tight_integration/check.exp CR-2005-11-25-1: Testsuite: o added some waits because of timing problems o fixed variable setting for issue 126 test Review: JG CR-2005-11-25-0: Testsuite: removed wait_for_enter call in checktree/functional/subordinate SG-2005-11-24-1: Bugfix: Reschedule of a parallel job crashes the qmaster Issue: 1916 Bugster: 6355263 Changed: qmaster Review: RD, JG CR-2005-11-24-1: Bugfix: a second started schedd/execd daemons may not stop if started when qmaster is down (enhancement for CR-2005-11-15-0) Bugtraq: 6349818 Issue: 1891 Changed: libgdi, execd Review: AA CR-2005-11-24-0: Testsuite: o fixed shadowd migrate test o removed which call at create_shell_script() Review: AA RD-2005-11-24-0: Bugfix: merge from s2_branch (RD-2005-11-07-1) qrsh -V doesn't always work. Issue: 1845 Bugtraq: 6346704 Review: AH RH-2005-11-24-0: Enhancm.: New arguments for scripts mk_catman and mk_ascman Changed: mk_catman and mk_ascman Review: AS SG-2005-11-24-0: Bugfix: fixed scheduler crash Changed: scheduler Review: RD DT-2005-11-23-2: Docs: o Fixed several man page issues. Issue: 1265, 1631, 1527, ?, ?, ?, ? Bugtraq: 6287860, 6287940, 6287910, 6353555, 4919544, 6314301, 4780562 Changed: Man pages DT-2005-11-23-1: Bugfix: o Fixed DRMAA Java language binding on HP Changed: aimk.site DT-2005-11-23-0: Cleanup: o Added missing SISSL header. o Created script to find env vars - Merged from DT-2005-11-22-0 (V60s2_BRANCH) Testsuite: o Fixed output/error files for drmaaj test Changed: testsuite, drmaaj CR-2005-11-23-0: Bugfix: o Fixed Insure memory leaks o Fixed wrong use of strncpy Review: AA, SG SG-2005-11-23-0: Bugfix: - mutually subordinating queues suspend eachother simultaneously - fixed the build for irix Issue: 1882 Bugster: 6354143 Review: JG Changed: scheduler SG-2005-11-23-0: Testsuite: added test for issue 1894 Review: JG Changed: checktree/functional/consumable_resources/ check.60.exp RH-2005-11-23-0: Testsuite: Required host for additional checktrees added Changed: testsuite Review: CR JG-2005-11-23-0: Bugfix: Fixed broken upgrade from 5.3 to 6.0. Review: MD EB-2005-11-22-0: Testsuite: create a testsuite test for issue 1874 Issue: 1911 Changed: testsuite Review: TBD CR-2005-11-22-0: Bugfix: o Fixed Insure memory leaks o removed obsolete start_commd variable o daemons now can be started daemonized with insure support Changed: libgdi, commlib, execd Review: AA RD-2005-11-22-0: Bugfix: default process priority on windows freezes the whole system until job is finished Changed: sge_shepherd Bugtraq: 6353638 Review: AH RD-2005-11-16-0: Testsuite: Fixed qtcsh test JG-2005-11-22-1: Testsuite: Use HP11 test binaries on HP1164 platform, if $SGE_ROOT/util/arch reports hp11 as binary platform. This is the case for our binary distribution. Review: CR JG-2005-11-22-0: Bugfix: libdrmaa.sl did not work on HP11 (link error) Issue: 1912 Bugtraq: 6354164 Review: SG AH-2005-11-22-1: Bugfix: Disable problematic schedd logging for jobs not finished as foreseen by reservation schedule Minor cleanup in spool lib error handling code that reads in sge_conf(5) and host_conf(5) Changed: scheduler Review: RD Bugtraq: 6353792 SG-2005-11-22-0: Bugfix: qstat incorrectly reports job scheduling failure Issue: 1894 Changed: scheduler Review: JG CR-2005-11-21-0: Testsuite: o enhanced issue 540 to support not NFS shared user home directory o fixed automatic install test to use default db and execd spool directory from manual installation o performance improvements for start_remote_prog and handle_vi_edit Review: MD JG-2005-11-18-1: Testsuite: Cleanup/fixes for bugs/issuezilla/1768 and bugs/issuezilla/1837. Added utility functions for handling of qtask file. Review: RD JG-2005-11-18-0: Testsuite: Fixed checks api/drmaa and bugs/issuezilla/1780. Review: CR CR-2005-11-18-0: Testsuite: fixed another problem with start_remote_prog Review: JG MD-2005-11-17-0: Bugfix: - installation of qmaster failed when using /etc/services Issue: 1900 Bugtraq: 6351728 Changed: util lib Review: CR CR-2005-11-17-2: Testsuite: - fixed tcl in issue test 1330 Review: RD CR-2005-11-17-1: Testsuite: - disabled file check of start_remote_prog - added file check to generated script Review: MD CR-2005-11-17-0: Testsuite: fixed unsupported warning message create for test checktree/bugs/issuezilla/1096/ Review: RD JG-2005-11-17-0: Bugfix: Fixed memory leaks in module test and qmaster (introduced during s2 development) Merge from V60s2_BRANCH, JG-2005-11-16-0 Review: CR RD-2005-11-16-0: Testsuite: Fixed qtcsh test Review: CR DT-2005-11-16-0: Bugfix: o Clearly explained -purge in qconf man page o Added -rsstree to qconf man page o Sorted options in qconf man page in alphabetical order Issue: 1783, 1897, 1898 Bugtraq: 6351174, 6351240, 6351278 Changed: man/man1/qconf.1 HP-2005-11-16-0: Bugfix: Fixed broken build on non-Solaris-non-Linux archs Review: EB Changed: shepherd, cull Makefile, ... CR-2005-11-16-0: Bugfix: o check file descriptor errors after select() errors in commlib Review: RD Changed: commlib DT-2005-11-15-2: Bugfix: o Clarified regex in complex man page o First pass at improving qconf -purge man entry o Uncommented -mhgrp in qconf man page o Fixed help output for qconf -purge Issue: 1782, 1783, 1872, 1895 Bugtraq: 6350714, N/A, 6347840, 6349351 Changed: man1, man5, qconf DT-2005-11-15-1: Testsuite: o Fixed test for IZ 1760 Changed: Test for IZ 1760 Review: pending DT-2005-11-15-0: Bugfix: o Fixed abort in drmaa_job_ps() Issue: 1892 Bugtraq: 6349972 Changed: JAPI Review: AH JG-2005-11-15-1: Testsuite: Fixed start of binaries (start_remote_prog) with user defined environment. Review: CR CR-2005-11-15-1: Testsuite: Fixed execd startup test Review: RD CR-2005-11-15-0: Bugfix: a second started schedd/execd daemons may not stop if started when qmaster is down (enhancement for CR-2005-11-14-1) Bugtraq: 6349818 Issue: 1891 Review: RD Changed: commlib, gdi, schedd, execd, qping JG-2005-11-15-0: Testsuite: Fixed invalid use of upvar with empty var name. Fixes and cleanup for checks bugs/issuezilla/1096, system_tests/clients/qdel, system_tests/clients/qstat, Review: CR RD-2005-11-14-0: Bugfix: qconf -ae hostname segfaults Changed: qconf Bugtraq: 6349608 Issue: 1884 Review: CR CR-2005-11-14-1: Bugfix: a second started schedd/execd daemons may not stop if started when qmaster is down Bugtraq: 6349818 Issue: 1891 Review: RD Changed: commlib, gd, schedd, execd, qping JG-2005-11-14-0: Bugfix: Update from 5.3 to 6.0 failed if the 5.3 cluster contained an empty complex. Bugtraq: 6349768 Issue: 1890 Review: MD RD-2005-11-14-0: Testsuite: - fixed autoinstallation spool dirs - fixed tests 1330, 1451, 1803 Review: CR CR-2005-11-14-0: Testsuite: new procedure check_all_system_times(): test sytem time on each cluster host before starting install_core_system. Review: RH DT-2005-11-11-0: Testsuite: o Added missing variable to Settings Bug: o Fixed problem with CR-2005-11-10-0 Changed: jUnit, commlib Review: CR CR-2005-11-11-1: Testsuite: o fixed incorrect change from CR-2005-11-11-0 CR-2005-11-11-0: Testsuite: o better remote file checking for start_remote_prog() Review: RD OJ-2005-11-11-0: Testsuite: Extended complex_global_host{} to test for regular expressions in complex values for test_restring; modified complex_run_job to account for issues found with qsub. Issue number: 461 DT-2005-11-10-0: Testsuite: o Fixed test for qsub -ac o Removed test for IZ1709 o Fixed version test in drmaaj o Expanded drmaa & drmaaj tests Changed: Testsuite, jUnit tests, DRMAA unit tests Review: pending CR-2005-11-10-3: Testsuite: fixed tcl error in system_tests/shepherd/job_start Review: RD CR-2005-11-10-2: Testsuite: fixed problem for start_remote_prog which and file test Review: RD CR-2005-11-10-1: Testsuite: fixed /system_tests/clients/qstat Review: RD MD-2005-11-10-0: Bugfix: merged from S2_BRANCH - nonadmin user installation of qmaster fails - SHADOW host install not able to install shadowd on more than one given shadowhost Bugtraq: 6347954 Issue: 1873 Changed: install_modules Review: RD EB-2005-11-07-0: Bugfix: unable to delete a configuration of a non existing host Bugtraq: 6319231 Issue: 1760 Changed: sge_qmaster, qconf Review: AH EB-2005-11-04-0: Bugfix: qmod -sq can kill a pe job in t state Bugtraq: 6287935 Issue: 1619 Changed: sge_qmaster, sge_execd Review: AH EB-2005-09-12-0: Cleanup: replaced fclose by FCLOSE in qmaster Issue: 1179, 185 Review: JG RD-2005-11-10-0: Bugfix: - sgeCA may be inconsistent after new installation - install script sets rsh_daemon to /usr/sbin/in.rshd on win32-x86 - qconf -mconf/-sconf does not resolve aliased host - lock initialisation for qsh Changed: install scripts, qconf, qsh Bugtraq: 6347351, 6348478 Issue: 1880 Review: MD CR-2005-11-10-0: Enhancem.: - added possibility to influence host name resolving by setting environment variables - added new commlib error: CL_RETVAL_CLOSE_ALIAS_FILE_FAILED Review: MD CR-2005-11-10-1: Testsuite: fixed bugs/issuezilla/1768 again Review: RD CR-2005-11-10-0: Testsuite: wait for pidfile on remote hosts in function get_grppid_of_job{} Review: JG OJ-2005-11-09-0: Testsuite: Created qtcsh test to test both in-line options and interactive options. Resides in checktree/system_tests/clients Changed: testsuite Issue: 419 Reviewed by: DT, JG DT-2005-11-09-1: Testsuite: o Hopefully fixed qsub -ac test Changed: qsub testsuite test DT-2005-11-09-0: Docs: o Cleaned up man pages o Removed -uall from qhold and qrls o Added list of ignored options for DRMAA Issue: 1870 Changed: man pages CR-2005-11-09-2: Testsuite: - Fixed problem for darwin hosts: ls doesn't return exit state != 0 when file is not existing, using test -f now - Fixed bugs/issuezilla/1768/ Review: AA, JG JG-2005-11-09-0: Testsuite: Fixes and cleanup in system_tests/execd/startup, system_tests/qmaster/reporting, system_tests/scheduler/soft_requests. Disabled test bugs/issuezilla/1760. Review: CR RH-2005-11-09-0: Testsuite: - Enhancements in the report procedures - Cleanup hooks for additional checktrees added Review: CR CR-2005-11-09-1: Testsuite: Fixes and cleanup for bugs/issuezilla/1760/ and /system_tests/clients/qconf/check.60.exp JG-2005-11-09-0: Testsuite: Fixes and cleanup for bugs/issuezilla/1768, bugs/issuezilla/1803, bugs/issuezilla/1848, module_tests/generic Review: CR CR-2005-11-09-0: Testsuite: Don't close connections which are ok after a failed test run. JG-2005-11-08-2: Testsuite: Fixed installation of Berkeley DB RPC server in non default cell. Fixed bugs/issuezilla/1803 and bugs/issuezilla/1848. JG-2005-11-08-1: Testsuite: Enhancements and cleanup in handle_vi_edit. Review: CR CR-2005-11-08-1: Testsuite: - enhanced shadowd migrate test. Test will now parse the qmaster messages file for qmaster exit value 100 state message Review: RD JG-2005-11-08-0: Testsuite: - fixed installation with bdb rpc server and default location for database directory. - some fixes and cleanup for qdel test - fixed deletion of jobs in reporting test Review: CR CR-2005-11-08-0: Testsuite: - added test for issue 1802 - increased cvs update timeout from 300 to 600 seconds Review: RH CR-2005-11-07-2: Testsuite: fixed wrong compile option setting from check-in CR-2005-11-07-0 ( when compile options are set to "none" ) CR-2005-11-07-1: Testsuite: fixed wrong compile option setting from check-in CR-2005-11-07-0 JG-2005-11-07-0: Bugfix: Connection between sge_qmaster and a Berkeley DB RPC server could timeout. Issue: 1866 Bugtraq: 6346696 Review: RD CR-2005-11-07-0: Testsuite: add -spool-classic compile option automatically when spool method "classic" is selected Review: RH RD-2005-11-07-0: Bugfix: qconf and qmaster accept invalid settings for queue complex_values Issue: 1847 Bugtraq: 6329832 Review: JG RC-2005-11-07-0: Enhance.: Tight SGE-SSH integration Issue: 1788 Changed: aimk, 3rdparty/remote/Makefile, sgessh.c Review: RD DT-2005-11-07-0: Testsuite: o Added test for issue 1760 o Added new operational locking routine o Fixed bugs in add_exechost Issue: 1839 Changed: Testsuite DT-2005-11-04-1: Bugfix: o Fixed issues from review of DT-2005-11-04-0 Issue: 1750 Bugtraq: 6250692 Changed: qmaster, config Review: OJ DT-2005-11-04-0: Bugfix: o Added separate reporting parameter for accounting flush interval o Fixed broken build on IRIX Issue: 1750 Bugtraq: 6250692 Changed: qmaster, config Review: AA CR-2005-11-04-0: Bugfix: additional cleanup/enhancement for CR-2005-11-04-0: - old qmaster will not spool spool data on shutdown when a shadowd qmaster has taken over Bugtraq: 6320869 Issue: 1820 Review: JG CR-2005-11-04-1: Bugfix: CSP consolidate error output if cert CA on client and server don't match Issue: 1802 Bugtraq: 6294875 Review: AA DT-2005-11-03-0: Testsuite: o Added test for reporting/IZ1750 Changed: system_tests/qmaster MD-2005-11-03-0: Bugfix: - autoinstall does not support csp installation mode - inst_sge -ux -host might be incomplete if not run from an admin host - Shadow master uninstallation cannot manage shadow_masters containing several lines Issue: 1861, 1451, 1578 Bugtraq: 6310168, 6239658, 6287828 Changed: inst_sge, install_modules Review: HP, CR RH-2005-11-03-0: Testsuite: Integration of additional checktrees implemented Changed: testsuite Review: CR RD-2005-11-03-0: Testsuite: fixed test for IZ 1803, 1780 and jobstart test Review: CR CR-2005-11-02-0: Bugfix: occasional "failed to deliver job" errors due to SIGPIPE in sge_execd Issue: 1856 Bugtraq: 6338314 Changed: commlib, execd Review: AA HP-2005-11-02-0: Bugfix: Fixed broken build Changed: libs/lck/Makefile SG-2005-10-31-0: Bugfix: - a sched conf change with a sharetree can result in a usage leak - sched conf changes master conf only when needed - minor perf. improv on job submit Issue: 1848 Bugtraq: 6342005 Changed: qmaster, scheduler, testsuite Review: AA RD-2005-10-28-0: Bugfix: fixed lock cleanup by SG-2005-10-25-0 lock was not initialized for non mt binaries Review: SG JG-2005-10-28-0: Testsuite: Prevent recursive calls of add_proc_error (when errors occur in error reporting) Fixed rlogin connection leaks. Cleanup. Review: CR OJ-2005-10-28-0: Testsuite: Extended qdel test to cover test cases described in IZ 125 Issue number: 125 Reviewed by: JG DT-2005-10-27-0: Bugfix: o Fixed handling of single-quotes Changed: sge_string.c Issue: 1762 Bugtraq: 6319233 Review: RD JG-2005-10-27-0: Testsuite: Avoid defunct processes after closing rlogin sessions. Changed: remote_procedures.tcl Review: RD DT-2005-10-26-0: Testsuite: o Improved test for qsub -ac for IZ1850 Changed: qsub test MD-2005-10-26-0: Bugfix: - typo in install script text - added autoinstall debug output - added a CLEAR between install dialogs Changed: inst_qmaster.sh, inst_berkeley.sh, inst_execd_uninst.sh DT-2005-10-25-2: Testsuite: o Improved test for qalter -[ads]c for IZ1762 Changed: qalter test DT-2005-10-25-1: Bugfix: o Fixed quoted qtask arg handing Issue: 1768, 1837 Bugtraq: 6317028, 6339756 Changed: sge_qtcsh.c Review: RD, OJ RD-2005-10-25-1: Bugfix: Fixed broken build on Solaris DT-2005-10-25-0: Testsuite: o Fixed init method for 1709 Changed: Test for IZ1709 SG-2005-10-25-0: Enhance.: - lock cleanup - schedd_config lock cleanup - sge_log lock cleanup - job startup message cleanup - schedd_lib cleanup, first part Changed: qmaster, scheduler, ... Review: JG RD-2005-10-25-0: Bugfix: - Binary jobs are problematic for starter and epilog scripts - wording with qconf -cq should be changed - testsuite does no recognize successfull tests Bugtraq: 6255111, 6207868 Issue: 1803, 1842 Review: HP, AH CR-2005-10-25-2: Cleanup: removed unused message macros CR-2005-10-25-1: Bugfix: fixed broken aimk clean option for some test binaries Review: RD CR-2005-10-25-0: Bugfix: added test_heartbeat.c which was missing by CR-2005-10-24-0 CR-2005-10-24-0: Bugfix: - sgemaster -migrate may not delete qmaster lock file and may break shadowd functionality - sge_qmaster daemon is running on both the master and shadow nodes after a long network failure - added test_heartbeat test binary - INFO text wasn't localized - testsuite test checktree/system_tests/shadowd/migrate - see also CR-2005-10-25-0 Bugtraq: 6333467, 6320869 Issue: 1828, 1820 Review: AH JG-2005-10-24-0: Bugfix: Fixed broken build on hp11. Review: CR AA-2005-10-24-0: Bugfix: - Changing the cwd flag in qmon - qalter has no effect Issue: 1838 Bugtraq: 6336519 Changed: qmon Review: SG CR-2005-10-21-0: Testsuite: fixed login problem on fast hosts for root login password question DT-2005-10-20-0: Bugfix: o Fixed NullPointerException in property setters and getters Issue: 1831 Changed: DRMAA-J Review: SG CR-2005-10-20-1: Testsuite: - testsuite performance tests (checktree/testsuite/check.60.exp) now tests all possible users on all possible hosts - new testsuite functions: file_procedures/print_xy_array() file_procedures/is_remote_file() Review: RH RD-2005-10-20-0: Bugfix: - shepherd doesn't handle qrlogin/qrsh jobs correctly - default PATH variable set for job insufficient for non-login shell jobs - fixed system_tests/shepherd/job_start Issue: 1772, 1695 Bugtraq: 6318018, 6288626 Changed: shepherd, execd Review: HP, CR CR-2005-10-20-0: Testsuite: fixed bug in testsuite checktree setup Review: RH DT-2005-10-19-0: Bugfix: o Fixed DRMAA memory leaks o Fixed handing of -b in DRMAA o Moved messages into header file Testsuite: o Added test for 1709 for C binding o Added test for 1770 o Clean up in TCL for Java language binding test Issue: 1763, 1770 Bugtraq: 6317048, 6320683 Changed: DRMAA Review: AH MD-2005-10-19-0: Bugfix: - added reprioritize param to default config - reprioritize man page description Changed: inst_qmaster.sh Review: SG, MD HP-2005-10-18-0: Bugfix: Fixed broken builds on aix43, irix65 and win32-x86 Changed: qtcsh, pdc, libsgeobj EB-2005-10-17-0: Bugfix: qconf -mattr can crash qmaster qconf -[dm]attr gets confused by shortcuts Issue: 1798, 1640 Bugtraq: 6292926, 6287946 Changed: sge_qmaster Review: SG DT-2005-10-14-0: Bugfix: o Fixed message number conflict Changed: msg_common.h CR-2005-10-14-2: Testsuite: enhancement: make a real install_core_system call when the first install (re-init active) fails after a check has failed CR-2005-10-14-1: Testsuite: fixed problem for additional checktrees config when creating a config from scratch CR-2005-10-13-1: Testsuite: added support for additional checktrees (better "change directory" search) MD-2005-10-13-0: Bugfix: sge qmaster startup and shutdown (non critical) error message as non root Issue: 1823 Bugtraq: 6252471 Review: CR CR-2005-10-13-0: Testsuite: added support for additional checktrees Review: RH (MD) SG-2005-10-13-0: Bugfix: configuring the halflife_decay_list crashes the qmaster Changed: qmaster, scheduler, qstat Issue: 1826 Bugtraq: 6333407 Review: RD RD-2005-10-12-0: Bugfix: fixed broken builds Review: MD SG-2005-10-12-0: Bugfix: Doing a qalter -l rsc=val on running jobs breaks consumable debitation Changed: sge_qmaster Issue: 1761 Bugtraq: 6315111 Review: RD MD-2005-10-10-1: Enhancem: changed inst_sge to provide a clear exit status to the testsuite Changed: inst_sge Review: HP RD-2005-10-10-1: Bugfix: - admin mail information is incorrect or queue error state setting does not work - tight integration - qrsh_exit_code file not written Changed: execd, shepherd Issue: 1680, 1679 Bugtraq: 6292751, 6292742 Review: HP RD-2005-10-10-0: Bugfix: gcc build on solaris Changed: aimk Review: JG MD-2005-10-10-0: Testsuite: added missing autoinst config line Changed: install_core_system/automatic/qmaster.60.tcl Review: RD RC-2005-10-08-1: Enhancem: support gcc4 as the build compiler (con't) Changed: 3rdparty/remote/rshd.c Review : pending RC-2005-10-08-0: Enhancem: PDC for AIX Changed : execd, shepherd Review : RD RC-2005-10-07-0: Enhancem: PDC for HP-UX Changed : execd, shepherd Review : AH RD-2005-10-05-0: Bugfix: - qstat -s p doesn't show pending array tasks while there are tasks of this job running - stat -U does not consider queue access for job and project access for queues - qstat -pe filter does not work - cull cleanup, removed XLISTDEF, ILISTDEF and SLISTDEF - fixed typo in JB_Type & inst_qmaster.sh Changed: inst_qmaster.sh, cull, qstat, qhost, qacct Issue: 1800, 1821, 1822 Bugtraq: 6301047, 6332876, 6332877 Review: EB CR-2005-10-05-0: Testsuite: make testsuite configuration more steady. Checking directory path names. Review: MD MD-2005-09-30-0: Testsuite: - used wrong local spooldirectory in autoinstall - adding exechosts to submithost list in autoinstall Changed: automatic/qmaster.60.tcl Review: CR CR-2005-09-29-0: Testsuite: - added new host configuration parameters: NFS-ROOT2NOBODY and NFS-ROOT2ROOT - added new global variable actual_user_system (install_core_system should use this to install an admin/normal/root user system) - removed parameter initialization for bdb_server, bdb_dir, spooling_method and cell to force this installation setup questions when running testsuite from scratch. - added additional optional parameter to handle_vi_edit (additional_expected_result4) - transform_date_time() had tcl error when the transform value was "-/-" (empty) - improved timing of remote procedures - do a qmod -c "*" in delete_all_jobs() for 60 systems - get_config() did not return error values - set_config() now supports also a change of spool directories - new optional parameter for is_job_running(): turn off job name parsing - shutdown_system_daemon(): improved shutdown - added new functions: get_string_value_between() get_uid() get_gid() get_execd_spooldir() get_file_uid() get_file_gid() switch_execd_spool_dir() submit_wait_type_job() - following functions should be implemented soon: switch_to_normal_user_system() switch_to_admin_user_system() switch_to_root_user_system() - added new test: /system_tests/shepherd/job_start Review: JG, HP (job_start_test) Issue: 1753 MD-2005-09-29-0: Bugfix: Using "-bup" with "-auto" breaks with later update release Issue: 1819 Bugtraq: 6314306 Changed: inst_backup.sh Review: CR JG-2005-09-29-0: Testsuite: Added documentation (ADOC) for use of Solaris zones in testsuite. Review: CR Issue: 1503 CR-2005-09-28-0: Bugfix: qping core dump with enabled message content dump Issue: 1810 Bugtraq: 6327427 Review: AA JG-2005-09-28-1: Testsuite: Added documentation (ADOC) for setup and cleanup functions. Review: CR Issue: 888 JG-2005-09-28-0: Testsuite: Added test system_tests/qmaster/complex. Cleanup. Issue: 117 Review: CR MD-2005-09-27-0: Testsuite: using a default dbd spooling dir, if not configured in testsuite config. Changed: autoinstall configfile creation Review: CR RD-2005-09-26-0: Bugfix: fstype does not recognize nfs4 share in all cases Issue: 1814 Bugtraq: 6328703 Review: JG MD-2005-09-23-1: Enhancem: - Added autoinstall support to testsuite - added test for Issue 1330, only possible with autoinstall support Changed: install_core_system Review: CR MD-2005-09-23-0: Enhancem: Check in inst_qmaster.sh, which I have forgotten with check in MD-2005-09-08-0 Changed: inst_qmaster.sh JG-2005-09-23-0: Testsuite: Added a testsuite test calling our module tests on all platforms. Fixed a number of compile errors in some module tests and made them part of a standard build. Review: RD CR-2005-09-23-1: Cleanup: updated comments in header files CR-2005-09-23-0: Bugfix: fixed duplicate message id entries RD-2005-09-23-1: Testsuite: qconf_issue1800 cleanup Review: JG CR-2005-09-22-2: Bugfix: confusing execd startup messages and delays in case of problems Testsuite: added test: system_tests/execd/startup Bugtraq: 6268799 Issue: 1801 Review: MD RD-2005-09-22-0: Bugfix: qstat -s p doesn't show pending array tasks while there are tasks of this job running Issue: 1800 Bugtraq: 6301047 Review: AH CR-2005-09-22-0: Testsuite: bugs/issuezilla/193: - test now is verifying if the remote execd is down after qconf -ke in scenario "issue193_deleted_job" CR-2005-09-22-0: Testsuite: new optional parameter for shutdown_system_daemon(): do_term_signal_kill_first: if 0 the procedure sends immediately a SIG_KILL signal SG-2005-09-22-0: Bugfix: - getting many E messages "failed building category string for job N" - fixed some module tests - module test for category generation - qmaster and common module tests are build per default now. Bugtraq: 2127607 Issue: 1652 Review: RD JG-2005-09-22-0: QA: Updated review template and doc. Review: RD RD-2005-09-21-0: Bugfix: Double error message Issue: 1767 Bugtraq: 6316995 Review: RH RD-2005-09-20-0: Bugfix: qacct -o -D output hard to parse Issue: 1686 Bugtraq: 6326150 Review: SG CR-2005-09-19-0: Testsuite: extended cvs update timeout JG-2005-09-17-0: Testsuite: Fixed check system_tests/config/path_alias. Review: CR RC-2005-09-17-0: Enhancem: support gcc4 as the build compiler JG-2005-09-15-0: Testsuite: Fixed checks functional/job_environment and system_tests/qmaster/monitoring. CR-2005-09-15-1: Testsuite: fixed commlib module test binaries test_virtual_event_client and test_virtual_qmaster. Testsuite test "commlib" failed because of long shutdown timeouts for virtual qmaster module test binary. CR-2005-09-15-0: Testsuite: modified testsuite commlib test to wait between single test runs SG-2005-09-15-0: Bugfix: fixed the qmaster calendar crash Issue: 1787 Bugtraq: 6322498 Review: RD SG-2005-09-14-0: Testsuite: fixed resource_consumable test RD-2005-09-14-0: Testsuite: fixed qmod_auto_reschedule test Review: CR SG-2005-09-14-0: Testsuite: added a qmaster monitoring test. Issue: 1765 JG-2005-09-14-0: Testsuite: Fixed check bugs/issuezilla/1780. RD-2005-09-13-1: Bugfix: qmaster keeps sending signals to execds Changed: qmaster Issue: 1665 Bugtraq: 6286510 Review: AH SG-2005-09-13-1: Bugfix: - fixed the timed event module test - fixed a script removel issue (performance) - fixed the los of tasks in system hold (memory leak) Issue: 1773 Bugtraq: 6318660 Review: JG SG-2005-09-13-0: Bugfix: second part of SG-2005-08-02-0, Tasks in hold state were not printed correctly and not covered in the schema Issue: 1733 Bugtraq: 6305095 Review: RD RD-2005-09-13-0: Bugfix: qstat output for alarm state reasons broken Changed qstat, qmon Issue: 1775 Bugtraq: 6323047 Review: CR AA-2005-09-12-0: Bugfix: ooops, removing /dev/null in sge_ca Changed: sge_ca Review: MD DT-2005-09-12-1: Bugfix: o Fixed quote processing Changed: sge_string.c Issue: 1556 Bugtraq: 6253860 Review: EB DT-2005-09-12-0: Testsuite: o Added test for issue 1556 DT-2005-09-09-0: Testsuite: o Reworked TS/jUnit integration to pass in env vars o Updated SessionImplJobTest to use new env vars o Added jUnit test for Issue 1770 o Updated drmaaj test to use drmaa-test.jar from source directory JG-2005-09-09-0: Bugfix: Fixed fix for IZ 1758, some cleanup. Fixed broken build on irix65. Changed: qsh (qrsh, qlogin) Issue: 1758 Bugtraq: 6313445 Review: DT MD-2005-09-08-1: Enhancem: - added some config parameters to autoinstall config file Changed: inst_template.conf MD-2005-09-08-0: Enhancem: - sge_ca auto enabled Changed: sge_ca Review: AA DT-2005-09-08-0: Testsuite: o Fixed test for Issue 1780 DT-2005-09-07-2: Bugfix: o Fixed broken build Changed: flatfile spooling DT-2005-09-07-1: Testsuite: o Added test for issue 1780 DT-2005-09-07-0: Bugfix: o Fixed broken build on IRIX 6.5 Changed: flatfile spooling EB-2005-09-05-0: Bugfix: eleminated static buffer in dstring implementation Issue: 265 Review: AH DT-2005-09-05-0: Bugfix: o Rewrote Flex parser for qconf o Fixed backslash line breaks in output from qconf o Added test cases o Cleaned up test comments Issue: 1749, 1780 Bugtraq: 6319228, 6256590 Changed: flatfile spooling Review: JG*2 AH-2005-09-05-0: Bugfix: aimk -spool-classic doesn't compile on linux/solaris Issue: 1764 Changed: all Review: JG AA-2005-09-05-0: Bugfix: sge_ca -usercert Bugtraq: 6318659 Issue: 1778 Changed: sge_ca Review: CR CR-2005-09-02-1: Bugfix: Merged CR-2005-09-02-1 from V60s2_BRANCH - fixed possible wrong errno usage Changed: commlib Review: AA CR-2005-09-02-0: Bugfix: Merged CR-2005-09-02-0 from V60s2_BRANCH - fixed memory leak in gethostbyaddr binary Changed: gethostbyaddr Review: AA CR-2005-09-01-1: Testsuite: Merged CR-2005-09-01-1 from V60s2_BRANCH - modified test_cl_commlib module test to support environment variable CL_RUNS which is used to stop the program execution after $CL_RUNS times. Changed: test_cl_commlib CR-2005-09-01-0: Bugfix: Merged CR-2005-09-01-0 from V60s2_BRANCH (- fixed message macro typos and output formats) Changed: qmod, qsub, qmaster, schedd, spoollib, utilib (all binaries have to be re-builded) Review: AA JG-2005-09-01-0: Testsuite: Fixed checks for SGE(EE) 5.3 (with CSP security): - functional/job_environment - functional/tight_integration - functional/usage - install_core_system - system_tests/clients/qalter - system_tests/clients/qdel - system_tests/qmaster/spooling MD-2005-08-31-1: Bugfix: auto installation doesn't provide sufficient diagnosis output Issue: 1330 Bugtraq: 6239653 Changed: autoinstall Review: CR CR-2005-08-30-0: Testsuite: added new command line switch to parse message macro dump files: check_macros DUMPFILE_A DUMPFILE_B IGNORE_BACKSLASH SG-2005-08-26-0: Cleanup: - fixed the build for irix 6.5 - enhanced RD-2005-08-16-0 (some global variables were missing) - extended the monitoring output Review: EB, (JG) Changed: qmaster, man-page SG-2005-08-25-0: Cleanup: Fixed build for linux 2.6 JG-2005-08-24-0: Testsuite: Removed debug code requiring user input during cluster installation. EB-2005-08-23-1: CVS Created V60s2_BRANCH --------------------------------------------------------- AA-2005-08-23-0: Bugfix: lFreeList missing Changed: all Review: RH EB-2005-08-23-0: Enhancem.: improved packing for qstat Changed: qmaster Review: (AH), SG RD-2005-08-19-0: Bugfix: Qrsh tries to free invalid pointer Changed: qrsh Issue: 1758 Bugtraq: 6313445 Review: SG SG-2005-08-20-1: Bugfix: qstat -j shows wrong message for parallel jobs which can't be dispatched Issue: 814 Bugtraq: 6287847 Review: RD SG-2005-08-20-0: Bugfix: - performance enhancement for pe-ranges and sequential soft requests - consumables as thresholds are not working correctly with pe jobs - Soft requirements on load values ignored - wrong soft requests decission - Testsuite test for consumables as load thresholds - Testsuite test for soft requests - Testsuite test for pe ranges Issue: 1626, 1738, 1742 Bugtraq: 6275789, 6306229, 6306834 Review: RD CR-2005-08-19-0: Bugfix: o fixed broken backup restore script Changed: dist/util/install_modules/inst_common.sh Review: RD DT-2005-08-18-0: Testsuite: o Added tests for backslash line continuation RD-2005-08-16-0: Cleanup: made the qmaster configuration thread safe Changed: all Review: SG, AA HP-2005-08-16-0: Bugfix: fixed NFS spooling problem Issue: 1752 Bugtraq: 6293411 Changed: daemons/common/err_trace.c, libs/uti/sge_uidgid.c, libs/uti/sge_uidgid.h review/HP-2005-08-10-0.txt Review: AH SG-2005-08-12-0: Bugfix: fixed the last, missing schema file Changed: message.xsd Review: DT HP-2005-08-16-0: Enhancem.: man page sge_conf now explains execd_paramd ENABLE_WINDOMACC Changed: sge_conf man page Bugtraq: 6294987 Review: AH DT-2005-08-12-0: Cleanup: o Added TODO comment for handling of architecture- specific reentrant functions in AH-2005-07-27-1. Changed: sge_hostname.c Review: SG CR-2005-08-12-2: Testsuite: - enhanced testsuite timeouts for drmaa test - modified test_drmaa module test binary to use local host for ST_INPUT_BECOMES_OUTPUT test - fixed problem for get_tmp_directory_name() and get_tmp_file_name() when the file isn't created before next call to tmp file name functions (functions returned the same file/directory name) Review: DT (drmaa) Changed: drmaa/check.60.exp drmaaj/check.60.exp file_procedures.tcl test_drmaa.c CR-2005-08-12-1: Enhancem.: - commlib will now check if cl_com_setup_commlib() is called twice (part 2/2) and also return with error when tread model is changed Review: RD Changed: cl_commlib.c MD-2005-08-11-1: Enhancem.: - merged windows installation from V60u4win_BRANCH (missed sgepasswd binary check, message output) - inst_common.sh MD-2005-08-11-0: Enhancem.: - merged windows installation from V60u4win_BRANCH Changed: - inst_sge, inst_execd.sh, inst_common.sh, inst_qmaster.sh CR-2005-08-10-0: Bugfix: - fixed broken test_drmaa module - enhanced commlib logging functions for unknown application threads - commlib will now check if cl_com_setup_commlib() is called twice (part 1/2) Changed: commlib gdi japi Review: AH, RD RD-2005-08-10-0: Bugfix: missing newline for qconf -mu RC-2005-08-08-0: Bugfix: wrong mem_total on MacOSX Issue: 1739 Bugtraq: 6307557 Changed: execd, loadcheck (lib/uti) Review: AH SG-2005-08-05-0: Cleanup: - Cull cleanup. The signature of the following functions has changed: - lListFree(**) - lFreeElem(**) - lAddList(*, **) - lRemoveElem(*, **) - lFreeWhat(**) - lFreeWhere(**) Changed: all Review: RD, AH AH-2005-07-27-1: Actually this should have been AH-2005-08-03 Bugfix: - DRMAA can abort in the middle of a session if NIS becomes unavailable - Fixed problems with non-reentrant parsing of -a datetime/-dl datetime - Removed all qmaster/libdrmaa link dependencies to non-reentrant C library functions for Solaris/Linux and added monitoring to make procedure to ensure qmaster/libdrmaa remain free of non-reentrant link dependencies Issue: 1727 Bugtraq: 6303671 Changed: all Review: CR, AA SG-2005-08-02-0: Bugfix: fixed the scheduler on linux Review: RD Changed: all RD-2005-08-01-0: Bugfix: - qmaster crashes with large number of qconf -aattr calls - qlogin -R does not work like documented - qstat -j doesn't print delimiter between jobs - qconf -as/-ah "" segfault Bugtraq: 6304466, 6304471, 6291023, 6304490 Issue: 1715, 1687, 1475, 1732 Review: SG SG-2005-08-02-0: Bugfix: qstat schema files are incomplete Issue: 1733 Bugtraq: 6305095 Review: LM (Sun Grid) AH-2005-08-02-0: Scripts: Added script source/scripts/nonreentrant.sh to search for non-reentrant linking dependencies in binaries, objects and libraries SG-2005-08-02-0: Enhancem.: enabled async gdi Changed: scheduler Review: AA SG-2005-08-01-1: Bugfix: bug in qstat -XML output (does not match the schema) Issue: 1731 Bugtraq: 6289455 Changed: qstat Review: RD SG-2005-08-01-0: Bugfix: strange reservation Issue: 1655 Bugtraq: 6287955 Changed: scheduler Review: RD JG-2005-07-29-0: Testsuite: check functional/consumable_resources erroneously reports a qhost error. CR-2005-07-29-1: Cleanup: code format cleanup CR-2005-07-29-0: Bugfix: - enabled multithreaded startup of scheduler daemon (Part 2) Fixed problems for linux 2.4 Kernel pthread library implementation - cleanup and TODO comments for sge_terminate() signal handler function - new parameter for get_conf_and_daemonize(): added an abort variable (modified by signal handler) - scheduler startup: commlib threads are now started as admin user - fixed core dump for multithreaded commlib clients trying to connect to a service which is unavailable. Review: SG Changed: sig_handlers.c execd scheduler commlib utilib gdi RD-2005-07-28-0: Bugfix: testsuite test issue 1473 failed qconf -sc didn't print a newline at end of file Review: CR JG-2005-07-28-2: Testsuite: check performance/throughput: - run with qmaster monitoring enabled - copy qmaster and schedd messages files to results dir cleanup JG-2005-07-28-1: Enhancem.: Distribution will contain additional Berkeley DB utilities (merged from V60u4win_BRANCH). Review: HP JG-2005-07-28-0: Bugfix: Fixed memory leaks (merged from V60u4win_BRANCH). Review: SG RD-2005-07-27-1: Cleanup messages cleanup Review: SG RD-2005-07-27-0: Bugfix: qhost -l for complexes broken Changed: qhost, qstat, schedd Bugtraq: 6287862 Issue: 1306 Review: SG JG-2005-07-27-0: Testsuite: Fixed testsuite problems on irix65. Issue: 1553 Improvements and cleanups in checks - bugs/issuezilla/126 - bugs/issuezilla/440 - functional/job_environment - system_tests/clients/qmod/general - system_tests/clients/qrsh AH-2005-07-27-0: Bugfix: Multi-threaded, multi-CPU username problems Lots of job parsing code wasn't MT safe Issue with IRIX compilation Issue: 1490 Bugtraq: 6242169 Changed: libdrmaa.so Review: JG SG-2005-07-27-0: Testsuite: added a test for resource reservation with array jobs. Changed: testsuite, test_drmaa Review: AH CR-2005-07-26-2: Bugfix: using of IP address as host names breaks unique hostname resolving (part 2) (additional handling for special host ip address 255.255.255.255) Issue: 1661 Bugtraq: 6282996 Changed: commlib Review: AH CR-2005-07-26-1: Bugfix: using of IP address as host names breaks unique hostname resolving (part 1) Issue: 1661 Bugtraq: 6282996 Changed: commlib Review: AH DT-2005-07-26-0: Testsuite: o Extended job environment functional test to test INHERIT_ENV and SET_LIB_PATH Issue: 1714, 1718 Bugtraq: 6297263, 6298056 CR-2005-07-26-0: Testsuite: added test for issue 1661 (ip address string as host name resolving) JG-2005-07-26-0: Testsuite: Error handling for setup and cleanup functions was broken. Issue: 1618 Cleanup. CR-2005-07-25-1: Enhancm.: - removed cl_commlib_cleanup_commlib() call from signal handler function sge_terminate() - disabled execd dispatcher leave on commlib errors (register at qmaster is also called inside dispatcher) - execd may loose dead children signal and may hence reaping of a child later (when next child is dead) - enabled multithreaded startup of scheduler daemon (Part 1) Review: SG Changed: commlib gdilib utilib execd schedd sig_handlers.c AH-2005-07-25-0: Bugfix: Changed execd h_rt from info to warning to improve diagnostics. Diagnost: Added logging if long delay with execd job spooling is encountered. Review: JG Issue: 483 CR-2005-07-25-0: Bugfix: qmaster crash on hostname resolution Issue: 1722 Bugtraq: pending Changed: utilib Review: SG JG-2005-07-25-0: Testsuite: Parsing of Gridengine messages: Search for linefeeds and tabs. RC-2005-07-23-0: Bugfix: several minor changes to get -umap compiling Review: Pending AH-2005-07-22-0: Bugfix: Added test_drmaa_perf utility that facilitates DRMAA performance testing Review: RH RD-2005-07-21-1: Bugfix: qrsh default job names must be consistent with limitations generally imposed on job names Changed: qsh Bugtraq: 6287865 Issue: 1363 Review: CR AH-2005-07-21-0: Bugfix: Use CL_RW_THREAD rather than CL_NO_THREAD for libdrmaa.so to fix problem of slow submission rate with drmaa_run_job() Review: CR Changed: libdrmaa.so Bugtraq: 6299982 Issue: 1724 SG-2005-07-21-0: Bugfix: no user notification or command hanging if an immediate job cannot be scheduled Review: JG, RD Changed: qmaster Bugtraq: 6298233 JG-2005-07-21-0: Testsuite: - the following checks failed, if first exec host was not the master host (has Solaris zones configured): o bugs/issuezilla/126 o system_tests/clients/qmod/general o system_tests/clients/qmake - extended bugs/issuezilla/126 to test issue on all architectures available in test cluster - check functional/migration failed if queue names were truncated by qstat CR-2005-07-21-0: Cleanup: shutdown of multithreaded commlib clients doesn't flush commlib internal message buffers Changed: commlib Review: AH RD-2005-07-21-0: Cleanup: unified message definitions part 3 Changed: sge_log.o Review: MD HP-2005-07-21-0: Bugfix: Nonexpressive error messages for SSL initialisation errors. Bugfix: qrsh failed when ARC not set and INHERIT_ENV=false set. Changed: sge_passwd.c, qlogin_starter.c Bugtraq: 6299351, 6299345 Review: AA JG-2005-07-20-0: Testsuite: Made check performance/throughput parse changed scheduler profiling output from SG-2005-07-20-0. SG-2005-07-20-0: Cleanup: - profiling is using the profiling output makro - profiling correctly outputs the number of send orders - monitoring is completly using dstrings now - added monitoring to the event master and timed event thread Changed: qmaster, scheduler Review: RD RD-2005-07-19-1: Bugfix: qacct -h shall not resolve hostnames Changed: qacct Bugtraq: 6295791 Issue: 1720 Review: CR AA-2005-07-19-0: Bugfix: make compiler happy Changed: sge_passwd Review: RD CR-2005-07-19-0: Cleanup: commlib: Performance Cleanup, etc. - If compiled with -DCL_DO_COMMLIB_DEBUG: o Use default commlib logging function o Fixed thread name setting for unknown appl. threads - Fixed wrong message send time in qping -dump output - optimized select() and poll() timeouts in tcp/ssl modules - Fixed division by zero exception in test_virtual_qmaster module Changed: commlib, test_virtual_qmaster Review: RD RD-2005-07-19-0: Cleanup: unified qstat output part 2 removed \n and \t from all message definitions modified all clients to print necessary newlines Changed: all Review: AA DT-2005-07-19-0: Bugfix: o Initialized INHERIT_ENV and SET_LIB_PATH to default values Changed: sge_conf.c Issue: 1718 Review: JG SG-2005-07-19-0: Bugfix: fixed the build for lx26-x86 Changed: sge_passwd Review: JG JG-2005-07-18-1: Testsuite: Check system_tests/shepherd/submit_del failed SGE 5.3 clusters in case of slow NFS or slow machines. JG-2005-07-18-0: Testsuite: Check bugs/issuezilla/540 failed in case of slow NFS MD-2005-07-18-0: Bugfix: typo in BT number Changed: Changelog SG-2005-07-15-0: Enhancem.: - made the scheduler configuration thread safe. - minor qmaster monitoring enhancements. Review: RD Changed: all HP-2005-07-14-0: Enhancem.: Merged sgepasswd functionality from V60U4WIN_BRANCH Changed: execd, shepherd, libs JG-2005-07-13-0: Testsuite: Fixed testsuite installation of SGE 5.3 clusters. SG-2005-07-13-1: Bugfix: Suspend on threshold not working for calendar disabled queues Review: EB Changed: scheduler Bugtraq: 6294052 Issue: 1691 RD-2005-07-13-1: Cleanup: removed unnecessary lib dependencies Review: JG SG-2005-07-13-0: Bugfix: . message cleanup - made modify requests atomar, when they were working on multiple objects. Changed: qmaster Review: RD RD-2005-07-13-0: Bugfix: unified qstat output part 1 removed some additional \t and \n in command line and xml output Changed: all Bugtraq: 6294118 Issue: 1692 Review: SG SG-2005-07-11-0: Bugfix: Array jobs sometimes get out of sync when grid master daemons is started/stopped Changed: qmaster, scheduler Bugtraq: 6295165 Issue: 1705 Review: EB, AH JG-2005-07-11-0: Testsuite: Fixed system_tests/clients/qmod/auto_reschedule. CR-2005-07-11-1: Bugfix: fixed thread monitor init Changed: utillib, execd Review: SG CR-2005-07-11-0: Testsuite: new test for issue #1502: check FD_SETSIZE DT-2005-07-11-0: Bugfix: o Fixed exception for invalid attributes in DRMAA-J o Fixed email property in DRMAA-J o Fixed memory leak in DRMAA-C o Cleaned up PartialTimestampFormatTest Changed: DRMAA Review: SG, AH Issue: 1650, 1651 Bugtraq: 6295231, 6295233 AS-2005-07-11-1 Cleanup: formatting changes to conform with rest of file Changed: dist/util/arch.dist RD-2005-07-11-1: Cleanup: changed full path to relativ path in commlib binaries no longer include full compile path Review: CR RD-2005-07-11-0: Bugfix: aix43 linker error after compiler update sge_qmaster Review: JG EB-2005-07-11-0: Enhancem.: cleanup and ADOC comments Changed: nothing Review: JG RC-2005-07-09-0: Enhancem.: install lx26 binaries without requiring to set SGE_ROOT Changed : arch.dist RD-2005-07-08-0: Bugfix: distinst does not set all file permissions Changed: distinst, mk_ascman, mk_catman Bugtraq: 6294632 Issue: 1697 Review: AS DT-2005-07-08-0: Enhancem.: o Added script to removed date tags from JavaDocs o Moved JavaDoc build dir to source/JAVADOCS Changed: build.xml, distinst SG-2005-07-08-1: Enhancem.: added monitoring capabilities to the qmaster Changed: qmaster, execd Review: JG SG-2005-07-08-0: Bugfix: removed mesage size limitation for qping Changed: commlib Review: CR SG-2005-07-07-0: Bugfix: fixed a memory leak and looking issue in the event master Changed: qmaster Review: JG JG-2005-07-07-0: Testsuite: configuration failed when hostlist (exec, shadow, submit only) contained an unknown host. Issue: 1594 Extended "fast_setup" to user config. CR-2005-07-07-1: Bugfix: - fixed memory leak in commlib for -DUSE_POLL compile definition - reviewd poll() implementation CR-2005-07-07-0: Cleanup: Merged qping man page changes from windows branch RD-2005-07-06-0: Cleanup: switched GDI_VERSION, GE_LONGNAME and GE_SHORTNAME from define to const char Changed: all Review: JG JG-2005-07-06-0: Testsuite: Fixed test system_tests/clients/qmod/general. Fixed adding exec hosts with Solaris zones to exec host list. RD-2005-07-05-1: Bugfix: wrong drmaa jnilib link on MacOS Changed: scripts/distinst Bugster: 6294397 Issue: 1694 Review: AH RD-2005-07-05-0: Bugfix: permission issues with noroot exported nfs shares Changed: inst_sge, inst_common.sh Review: MD RD-2005-07-04-0: Bugfix: qrsh returned exit status 1 instead of 0 Changed: shepherd Review: CR RC-2005-07-03-1: Bugfix: bug in RestartService() Changed: darwin_template Review: Pending RC-2005-07-03-0: Enhancem.: use poll() instead select() (enable with -DUSE_POLL) Changed: cl_tcp_framework.c, cl_ssl_framework.c Issue: 1510 Review: CR CR-2005-07-01-1: Cleanup: commlib: optimized host compare function Review: RD CR-2005-07-01-0: Testsuite: added missing translation for rpc server installation script Review: RD RD-2005-07-01-0: Bugfix: distinst ignored sgepasswd and tried to copy libXltree on win32-x86 sgepasswd not covered by setfileperm.sh Changed: setfileperm.sh, distinst Review: JG, CR CR-2005-06-30-0: Cleanup: commlib: - new logging for commlib debuging with definition of CL_DO_COMMLIB_DEBUG at compile time - optimized locking when receiving messages Review: RD RD-2005-06-30-2: Bugfix: disabled admin mail if qrsh_exit_code can't be read removed empty lines in shepherd trace file Review: JG RD-2005-06-30-1: Testsuite: fix for running as other user as sgetest Review: CR RD-2005-06-30-0: Bugfix: backup and restore was broken several comments corrected Changed: inst_common.sh Review: JG JG-2005-06-30-0: Testsuite: Check for core files during cluster shutdown failed if non local spool directories were used. Fixed incomplete cleanup of consumable resources test. New testsuite option "fast_setup" to skip expensive verifications of the configuration file. Review: CR AA-2005-06-28-0: Bugfix: unmanage qsort args field when not compiled with SGE_PQS_API Changed: qmon Review: CR CR-2005-06-28-0: Cleanup: commlib: skipped unnecessary string compares when parsing message information headers Review: AA CR-2005-06-27-1: Enhancem.: - added send message queue for connection handles (performance enhancement) - code cleanup Review: RD CR-2005-06-27-0: Testsuite: throughput test: using tcl timestamp command instead of clock seconds command which caused time measurement errors on some architectures Review: RD RD_2005-06-27-0: Bugfix: installation on Red Hat creates wrong rc*.d script names Bugster: 6277874 Issue: 1669 Review: CR JG-2005-06-27-0: Testsuite: Closing rlogin connections sporadically failed. Review: CR RD-2005-06-24-1: Bugfix: fixed broken build on linux RD-2005-06-24-0: Bugfix: tight integration: temporary files are not deleted at task exit Changed: shepherd Bugster: 6287867 Issue: 1410 Review: JG CR-2005-06-23-1: Cleanup: commlib: - substates for CL_CLOSE state was not initialized - round robin connection handling for writeing Review: RD CR-2005-06-23-0: Testsuite: Fixed possible NFS problem when waiting for certificates tar file at CSP execd installation RD-2005-06-21-2: Bugfix: several install script fixes Bugtraq: 5063313, 5090187, 5056331, 6289240 Issue: 1084, 1248, 1149, 1249 Review: MD CR-2005-06-22-0: Enhancem.: - added received message queue for connection handles (performance enhancement) - code cleanup Review: RD AH-2005-06-22-1: Bugfix no event client known to process acks error logging Issue: 1675 Bugtraq: 6289273 Changed: sge_qmaster Review: JG, EB AH-2005-06-21-1: Bugfix: Jobs submitted with -v PATH do not retain $TMPDIR prefixed by Grid Engine as required for tight integration Issue: 1674 Bugtraq: 6288588 Changed: sge_execd Review: RD RD-2005-06-21-1: Bugfix: sge_shepherd SEGV's when it tries to fopen the usage file Bugtraq: 6288156 Issue: 1673 Review: AH RD-2005-06-21-0: Bugfix: suspend not working under Mac OS X Bugtraq: 6287958 Issue: 1671 Review: AH CR-2005-06-20-1: Enhancm.: - commlib thread list: o added new parameter to specify cleanup function o added new parameter to specify user data - enhanced commlib debuging when CL_DO_COMMLIB_DEBUG is defined at compile time Review: AA RD-2005-06-20-0: Bugfix: serveral install/uninstall issues Bugtraq: 6287831, 6287824, 6287850, 6245812, 6287826, 6287917 Issue: 1617, 1481, 1028, 1108, 1528, 1670 Review: MD CR-2005-06-20-0: Cleanup: - re-added secure dummy string (to be able to check files for correct compile mode in install script) - fixed wrong variable type Review: JG EB-2005-06-17-0: Bugfix: delivery of queue based signals to execd repeated endlessly Issue: 1665 Bugtraq: 6286510 Review: AH SG-2005-06-17-0: Bugfix: made SG-2005-06-15-1 more efficent Changed: master RD-2005-06-17-0: Bugfix: autoinst execd uninstallation disables and suspends whole queues instead of instances only introduced with RD-2005-05-23-0 Review: MD AH-2005-06-16-3: Bugfix: tru64 binaries use 64-bit u_long as u_long32 Issue: 1399 AH-2005-06-16-3: Bugfix: qtcsh will not compiles on Fedora Core 2 or later Issue: 1571 RD-2005-06-16-0: Bugfix: qconf -Xattr does not resolve fqdn hostnames Issue: 1667 Bugtraq: 6285898 Review: EB AH-2005-06-16-0: Bugfix: job wallclock monitoring and enforcement considers prolog/epilog runtime part of net job runtime Fixed popular "allready" typo Issue: 1666 Bugtraq: 6286533 Review: HP JG-2005-06-16-0: Testsuite: Improved output of performance/throughput test. Fixed handling of pe and ckpt objects with long queue lists in 5.3 systems. SG-2005-06-15-1: Bugfix: resource allocation shown by qstat/qhost not consistent with resource utilization Changed: qmaster Issue: 1658 Bugtraq: 6281440 Review: EB JG-2005-06-15-0: Cleanup: enabled more warnings for compilation with -Werror on irix65. Bugfix: - fixed broken build for lx24-alpha - fixed a memory leak Changed: all Review: RD RD-2005-06-15-0: Bugfix: GEMM doesn't like behavior of checkin RD-2005-05-23-0 Review: MD SG-2005-06-15-0: Bugfix: - fixed a couple performance bugs - fixed the build for irix - fixed unknown event handler message on qmaster startup Changed: qmaster, scheduler Review: RD AH-2005-06-14-0: Bugfix: Overhead with job execution could lead to overoptimistic backfilling and break resource reservation Issue: 1662 Bugtraq: 6283308 Review: AS, CR HP-2005-06-14-0: Enhancm.: All sgepasswd related functionality merged from V60U4WIN_BRANCH Review: MD Changed: execd, shepherd, libs/uti, libs/sgeobj, libs/gdi, utilbin/... SD-2005-06-13-0: Enhancm.: Added Parallel Environment Queue Sort API as compile-time option (-DSGE_PQS_API). RD-2005-06-13-0: Bugfix: fixed broken sge_passwd Review: AA JG-2005-06-13-0: Bugfix: Fixed missing initialization. FES_Type was missing in cull namevector. Changed: all Review: CR AA-2005-06-10-3: Bugfix: html man page generation scripts mismatched man pages with same name (sgepasswd (1,5)), added -s section to man -> restricts generation of man pages to solaris man Changed: scripts/mk_ascman AA-2005-06-10-2: Bugfix: merge from V60u4win - removal of link dependency to libspoolc.so Issue: 1569 Review: JG Changed: qmon AA-2005-06-10-1: Bugfix: merge from V60u4win - sgepasswd, static linking of libcrypto Changed: sgepasswd, aimk Review: EB AA-2005-06-10-0: Bugfix: merge from V60u4win - qmon basic test, exechost consumable/fixed attribute removal on x86 and sol-x86 architecture Changed: qmon Review: CR RD-2005-06-10-0: Enhancem: add rpm switch to mk_dist Changed: mk_dist Review: AA JG-2005-06-08-0: Bugfix: qalter -p shows wrong priority. Issue: 403 Bugtraq: 2125401, 4769608 Changed: sge_qmaster Review: CR AH-2005-06-08-0: Bugfix: drmaa_exit() and qsub -sync y causes qmaster error logging if host is not admin host Issue: 1659 Bugtraq: 6279402 Changed: sge_qmaster Review: CR JG-2005-06-07-1: Bugfix: qmaster profiling can only be turned on by restarting qmaster. Issue: 1601 Bugtraq: 6281462 Changed: sge_qmaster Review: SG JG-2005-06-07-0: Testsuite: New check system_tests/qmaster/profiling. Minor fixes for tests functional/usage and functional/tight_integration. RD-2005-06-07-0: Bugfix: backup script does not backup sgeCA directory for CSP systems Changed: inst_common.sh Bugtraq: 6262009 Issue: 1591 Review: CR CR-2005-06-07-0: Testsuite: o test_issues module test now raise memory allocation size slower (some machines started swapping so that commlib/issues test failed o raised timeout waiting for output in commlib/issues test o raised timeout waiting for file in bugs/issuezilla/1489 test AA-2005-06-06-0: Testsuite: comment caused problem for newer expect version EB-2005-06-06-0: Bugfix: file "locking" for sgepasswd Changed: sgepasswd Review: HP SG-2005-06-06-0: Bugfix: - minor performance improvement in the qmaster, scriptfile was unlinked twice SG-2005-06-06-1: Testsuite: new resource reservation test Issue: 1488, 697, 714, 703 Review: JG AH-2005-06-06-0: Bugfix: Race condition with qsub -sync yes and drmaa_wait() if job exits directly after being submitted Issue: 1657 Bugtraq: 6273217 Changed: qsub, libdrmaa.so Review: DT SG-2005-06-06-0: Bugfix: - minor performance improvement in the qmaster, scriptfile was unlinked twice - Resource filtering with qhost broken Changed: qmaster, scheduler, qhost Issue: 1531 Bugtraq: 6280698 Review: JG RD-2005-06-02-3: Bugfix: upgrade from 5.3 -> 6.0 add managers to operators Changed: install scripts Review: MD RD-2005-06-02-2: Bugfix: qstat -xml -urg output contains badly formatted numbers Changed: qstat Bugtraq: 6278727 Issue: 1656 Review: AH RD-2005-06-02-1: Bugfix: inst_sge -sm don't install a startup script enh. MacOS Startup Item BDB RPC Server Changed: install scripts Bugtraq: 6278140 Issue: 1646 Review: MD AA-2005-06-02-0: Testsuite: Bugfix: - shadowd install send answer for install as admin user MD-2005-06-02-0: Bugfix: Merged from V60u4win_BRANCH - shadow_masters file could not be created on user root=nobody mounted fs - settings file must be sourced for shadow host installation. Changed: inst_sge Review: HP SG-2005-06-02-0: Bugfix: qconf -tsm command never stops writing output Changed: scheduler Bugtraq: 6279409 Issue: 1616 Review: JG RD-2005-06-02-0: Bugfix: Close integration cause jobscripts with multiple mprun commands to be killed. Changed: shepherd Bugtraq: 6269411 Issue: 1654 Review: AA AH-2005-06-01-3: Bugfix: libdrmaa.so crashed if monitoring was enabled Review: JG AH-2005-06-01-1: Bugfix: drmaa_job_ps() returns DRMAA_PS_QUEUED_ACTIVE for finished array job rather than DRMAA_PS_DONE Issue: 1648 Bugtraq: 6278147 Changed: libdrmaa.so Review: DT RD-2005-06-01-0: Bugfix: inst_sge -db error on MacOS Changed: install scripts Review: MD Issue: 1647 Bugtraq: 6278146 JG-2005-05-31-0: Testsuite: Bugfixes: - do not search for core files in non existing dir - do not try hostname resolving on hostgroups SG-2005-05-30-0: Bugfix: qconf -mq coredumps Changed: qconf Review: DT, JG, RD Issue: 1642 Bugtraq: 6277909 AH-2005-05-30-1: Bugfix: Removed outdated and unused DIST_VERSION in inst_sge Review: MD AH-2005-05-27-1: Doc: Job error state is not documented in the qstat man page Issue: 1631 AH-2005-05-27-0: Bugfix: Problem with large job output files under 32-bit Linux Review: CR Issue: 1628 Bugtraq: 6277104 SG-2005-05-27-0: Bugfix: fix qstat -t again (SG-2005-05-25-0 was not enough) Review: CR Changed: qstat Issue: 1420 Bugtraq: 6218877 JG-2005-05-27-0: Testsuite: - Spooldir of first exec host was not cleared during installation. - If multiple exec hosts used the same spool directory, spool directories of previously installed exec hosts were deleted in subsequent exec host installations. - Check for core files in the spool directories after cluster shutdown. - Some cleanup. DT-2005-05-25-0: Bugfix: o Enforced thread safety in DRMAA Review: SG Changed: DRMAA Issue: 1632 Bugtraq: 6267238 SG-2005-05-25-0: Bugfix: fix qstat -t again (SG-2005-01-21-0 was not enough) Review: EB Changed: qstat Issue: 1420 Bugtraq: 6218877 AH-2005-05-25-0: Bugfix: Fixed Makefile insufficiencies Doc: Added scheduler comments AA-2005-05-24-1: Bugfix: usage output Changed: qmon AA-2005-05-24-0: Bugfix: FAQ incorrectly states glib 2.1.3 is required for Linux Issue: 410 Changed: www Review: AS JG-2005-05-24-0: Testsuite: Fixed installation of SGE 5.3 as non root user. SG-2005-05-23-1: Bugfix: job_load_adjustements is not correctly working when parallel jobs are submitted. Changed: scheduler Issue: 1620 Bugtraq: 6268707 Review: EB AH-2005-05-23-0: Example: Added compile.c for distributed aimk Review: JG AA-2005-05-23-1: Bugfix: Have the Queue Instances tab within Qmon display queues in alphabetical order (like 5.3 did with the queue tab) Issue: 1602 Changed: qmon Review: CR AA-2005-05-23-0: Bugfix: qmon kills a system Bugtraq: 6274467 Issue: 1446 Changed: qmon Review: CR RD-2005-05-23-0: Bugfix: execd auto_install performance bottleneck Changed: install modules Bugster: 6272451 Issue: 1627 Review: MD SG-2005-05-22-0: Bugfix: test_drmaa seg faulted on linux. Changed: test_drmaa Review: JG AA-2005-05-20-1: Bugfix: qmon: complex attributes not removeable Bugtraq: 6252525 Issue: 1625 Changed: qmon Review: RD AA-2005-05-20-0: Bugfix: qmon crash (segmentation fault) on Solaris64 Bugtraq: 6250603 Issue: 1541 Changed: qmon Review: CR EB-2005-05-20-0: Bugfix: Performance prob. with qconf -mattr exechost global Bugtraq: 6266392 Issue: 1623 Changed: sge_qmaster Review: AH RD-2005-05-20-0: Bugfix: performace bottleneck with subordinate list BugTraq: 6266450 Issue: 1624 Changed: qmaster Review: SG SG-2005-05-12-2: Bugfix: missleading qstat -j messages in case of rr BugTraq: 6252469 Issue: 1547 Changed: scheduler Review: RD RD-2005-05-19-1: Bugfix: qstat -j "" results in a segmentation fault BugTraq: 6273006 Issue: 1610 Changed: qstat Review: AH RD-2005-05-19-0: Bugfix: qrsh/qsh/qlogin reject -js option BugTraq: 6269305 Issue: 1621 Changed: qsh, qrsh, qlogin Review: AH SG-2005-05-19-1: Bugfix: incomplete resource reservation with array jobs Changed: scheduler Review: RD, AH BugTraq: 6260656 Issue: 1583 SG-2005-05-19-0: Bugfix: Wildcards in PE Name Cause Unusual Behavior Changed: scheduler Review: AH BugTraq: 6265154 Issue: 1597 AH-2005-05-17-0: Doc: Fixed inconsistent naming for "weight_waiting_time" in sge_priority(5) BugTraq: 6269033 Changed: man CR-2005-05-13-6: Bugfix: commlib acknowledge message should be send when a acknowledged message arrives and not when gdi removes the message from commlib buffer. Review: pending Changed: commlib CR-2005-05-13-5: Enhancm.: added DPRINTF() statements where the act_qmaster file is read JG-2005-05-13-0: Bugfix: fixed broken build on tru64 CR-2005-05-13-4: Bugfix: - fixed broken build for irix65 and hp11 Changed: commlib CR-2005-05-13-3: Bugfix: fixed bug in commlib Makefile Changed: source/libs/comm/Makefile CR-2005-05-13-2: Bugfix: new message will trigger write thread only when connection is in state CL_CONNECTED or CL_CLOSING. This will fix unnecessary high write thread cpu usage when connection is not yet established. Review: AA Changed: commlib (ibs/comm/cl_communication.c) CR-2005-05-13-1: Testsuite: - fixed commlib throughput test Changed: testsuite CR-2005-05-13-0: Cleanup: - cleanup of some commlib message logging and added new error messages Changed: commlib CR-2005-05-12-0: Enhancm.: - Added new cvs directory: source/libs/comm/examples - added 2 commlib demo modules and Makefile: client_demo and server_demo Changed: Makefile of commlib, source/scripts/zerodepend RD-2005-05-12-0: Bugfix: fixed broken build on HP11 no dlopen on HP11 Changed: sge_passwd, shepherd Review: TBD SG-2005-05-11-2: Bugfix: fixed a wrong flush settings in the scheduler event client Changed: scheduler Review: pending CR-2005-05-11-1: Enhancm.: added prototype for new module test: test_thread_throughput CR-2005-05-11-0: Bugfix: high CPU load of qmaster even on empty cluster Issue: 1612 Bugtraq: 6267932 Changed: commlib Review: JG, AS JG-2005-05-11-0: Testsuite: Improved error messages and error handling when opening rlogin connections (open_remote_spawn_process). Improved/fixed test performance/throughput: - more finegrained process monitoring - use a local filesystem as working directory for job submission and execution to avoid NFS overload. Fixed pe/ckpt to queue association for SGE 5.3. Review: CR SG-2005-05-11-0: Bugfix: - fixed delay in event client after recieving an event package. - sge_passwd did not build on irix Changed: event_client, sge_passwd Review: CR EB-2005-05-11-0: Bugfix: renamed sge_passwd man pages to sgepasswd RC-2005-05-11-0: Enhancem.: install MacOSX Startup Item Review: RD Changed: dist/util/install_modules/inst_common.sh Added: dist/util/rctemplates/darwin_template JG-2005-05-10-0: Testsuite: Fixed tests to run testsuite with SGE(EE) 5.3: - bugs/issuezilla/423 - bugs/issuezilla/1126 (N1GE6 only test) - system_tests/clients/qalter - system_tests/clients/qmod/reschedule - system_tests/config/file_parsing - system_tests/config/submit_hosts SG-2005-05-10-0: Bugfix: fixed the flushing in the event master Review: pending Changed: qmaster EB-2005-05-10-0: Bugfix: Domain admin <-> domain user problem (windows) Bugtraq: 6239470 Review: HP HP-2005-05-09-1: Bugfix: Excluded qmon, sge_master, sge_shadowd and sge_scheddd install from distinst for Windows Changed: scripts/distinst CR-2005-05-09-1: Bugfix: Repeated logging of the same message produces giant logging files Issue: 1598 Bugtraq: 6267245 Changed: gdi library (all binaries) Review: AH HP-2005-05-09-0: Bugfix: Excluded DB and DRMAA install from distinst for Windows Changed: scripts/distinst CR-2005-05-09-0: Testsuite: Fixed bug in check_local_spool_directories() JG-2005-05-06-2: Bugfix: Fixed typo in qstat error message. Issue: 1580 Review: RD JG-2005-05-06-1: Testsuite: check functional/backup_restore only applies to Grid Engine version 6.x. JG-2005-05-06-0: Testsuite: Make opening of rlogin connections safer and faster. Fixed check bugs/issuezilla/403. CR-2005-05-06-0: Testsuite: - fixed module test: test_cl_commlib_client - reworked handle_vi_edit() - fixed bug in resolve_queue() - fixed bug in create_shell_script() JG-2005-05-04-0: Cleanup: enable compilation with -Werror on irix65. Review: CR, RD CR-2005-05-04-0: Testsuite: - using shorter queue names. Long queue instances names break qstat parsing procedures() - new procedure: resolve_queue() - added queue resolving for add_queue() and get_queue_state() CR-2005-05-03-4: Testsuite: install_core_system now supports load sensor installation on nosuid root mounted disks, if qmaster is installed on fileserver. CR-2005-05-03-3: Testsuite: fixed bug in create_gnuplot_xy_gif() which had the effect that png terminal was used even when gif terminal is supported. (bug is from checking CR-2005-04-28-1) CR-2005-05-03-2: Testsuite: modified module test for virtual event/gdi clients to shutdown when not able to connect to virtual qmaster for more that 15 min. DT-2005-05-03-0: Bugfix: o Fixed drmaa_control(ALL) Cleanup: o Removed redundant function declaration Changed: JAPI, cull_parse_util.h Review: AH Issue: 1595 Bugtraq: 6264592 CR-2005-05-03-1: Testsuite: - added support for new message macro format: sge_X32CFormat and sge_U32CFormat CR-2005-05-03-0: Testsuite: - moved some functions from main testsuite script to tcl_files directory - added new configuration paramter: mail_application (user can now select mailx, sendmail or a mail wrapper script) - /bugs/issuezilla/403: using string match to compare output of qalter - removed not active code added some output - fixed path_alias_test - new optional parameter for create_shell_script(): set without_start_output 1 will not print out start/end message - testmails now contain testsuite configuration - setting default resolve_host() resolving to long RD-2005-05-03-0: Cleanup: compile on solaris 10 basic types renamed Changes: all Issue: 1514 Bugtraq: 6264567 Review: AA AH-2005-05-02-0: Bugfix: qrsh shouldn't block SIGTSTP Issue: 1464 Review: NJ Changes: qsh, qrsh, qlogin AS-2005-05-02-0: CVS Created V60u4win_{TAG,BRANCH} --------------------------------------------------------- CR-2005-05-02-1: Testsuite: o Fixed bug in testsuite test system_tests/clients/qconf: - message parsing for long host names was wrong - check did not re-add the execd on failure - check did unset the SGE_SINGLE_LINE environment variable which is already set from testsuite main CR-2005-05-02-0: Testsuite: o Fixed tcl error in throughput test CR-2005-04-29-0: Testsuite: o Fixed resolve_host(): - There was no cache for the optional long resolving parameter o Fixed set_cqueue_specific_values(): (for 6.0 GE) - procedure removed "=" from cluster queue instances: e.g.: "fangorn=test2=4" was modified to "fangorn=test2 4" - procedure saved host specific entries without resolving the names DT-2005-04-29-0: Testsuite: o Added junit test for issue 1595 - Note: 1595 hasn't been fixed yet, so the test will fail JG-2005-04-29-0: Bugfix: Fixed broken build. CR-2005-04-28-1: Testsuite: o create_gnuplot_xy_gif() now uses png terminal when gnuplot is not supporting gif terminal CR-2005-04-28-0: Testsuite: o replaced incr with expr, because for tcl 8.4/ expect 5.40.0 there is a problem with [clock -milliseconds] integer value (which is negative) CR-2005-04-27-5: Testsuite: o performance cleanup o added new test to test testsuite performance o added debug messages DT-2005-04-27-0: Experim.: o Added new "gridless" DRMAA Java implementation to experimental packages CR-2005-04-27-4: Testsuite: added a testsuite test template: checktree/test_template/check.60.exp CR-2005-04-27-3: Testsuite: Enhancement: speedup in testsuite command execution CR-2005-04-27-2: Testsuite: modified testsuite menu to support subdirectory names CR-2005-04-27-1: Testsuite: modified test bugs/issuezilla/1489: - using testsuite standard procedures to delete files - increased file wait timeout because NFS may cause timing problems RH-2005-04-27-0: Testsuite: added new test system_tests/scheduler/urgency Review: JG CR-2005-04-27-0: Testsuite: modified test bugs/issuezilla/1334: - using testsuite standard procedures to delete files and start programs - new optional parameter for delete_file() do_wait_for_file: waiting for file to appear is on (default), but can be switched off JG-2005-04-26-0: Testsuite: Fixed testsuite startup for SGE(EE) 5.3 clusters. Added scripts to list all / verified tests. Some cleanup, fixed typos. CR-2005-04-26-0: Testsuite: enhanced functional/consumable_resources test to install 2 load sensors AA-2005-04-25-0: Bugfix: Can't select 'slots' in select box when adding consumables for execution host Changed: qmon Bugtraq: 6260729 Issue: 1586 Review: AH DT-2005-04-25-0: Bugfix: o Repaired japi_exit() locking Changed: JAPI Review: SG, AH Bugtraq: 6260704 RD-2005-04-25_1: Bugfix inst_sge backup and restore issues shadow_masters not backuped running qmaster not recognized Bugster: 6259993 Issue: 1585 Review: MD HP-2005-04-25-0: Bugfix: Fixed MANPATH for Interix Changed: dist/util/arch dist/util/arch.dist Review: AS CR-2005-04-25-1: Testsuite: moved system_tests/geee/sharetree/check.exp to system_tests/scheduler/sharetree/check.exp CR-2005-04-25-0: Testsuite: fixed backup_restore test for rpc spooling Review: RD RH-2005-04-25-0: Testsuite: added new test: system_tests/scheduler/deadline Review: JG RD-2005-04-25-0: Bugfix: dbwriter install failed if clause used undefined variable Changed: dist/util/inst_qmaster.sh Review: MD AS-2005-04-25-0: Bugfix: changed base directory of Windows man pages to /catman/cat Changed: scripts/distinst dist/util/arch MD-2005-04-22-0: Bugfix: - adminrun didn't switch to adminuser under windows - global execd spooldir was not created during execd-install Changed: adminrun.c, inst_execd.sh Review: HP AA-2005-04-22-0: Bugfix: qmon cluster queue modify cancel working incorrect Bugtraq: 6260024 Issue: 1582 Changed: qmon CR-2005-04-22-0: Testsuite: fixed test: system_tests/scheduler/tickets - test did not wait till all tickets are calculated from scheduler - new optional parameter for parse_qstat() and get_extended_job_info(): do_replace_NA will show NA for not availabe CPU load instead of 0 Review: pending RD-2005-04-22-0: Testsuite: - functional/backup_restore enhancement works now with classic, local and rpc spooling Review: CR HP-2005-04-22-1: Bugfix: Added check for network path in execd-rc-script Changed: dist/util/rctemplates/sgeexecd_template Review: MD HP-2005-04-22-0: Bugfix: Added correct include path Changed: libs/uti/sge_unistd.h AS-2005-04-22-0 Bugfix: fixed common package content Changed: source/scripts/mk_dist HP-2005-04-21-0: Bugfix: Created SGE_STAT Macro for Interix Changed: libs/uti/sge_unistd.h RD-2005-04-21-0: Testsuite: - added new test: functional/backup_restore/ creates backup, removes cell directory and restores from backup. Changed: Testsuite Review: CR SG-2005-04-21-0: Bugfix: potential qmaster sec. fault. Changed: qmaster Issue: 1579 Bugtraq: 6259380 Review: DT, JG RH-2005-04-21-0: Testsuite: - added new test: system_tests/scheduler/tickets Checks the calculation of overwrite and functional ticket, the start order and the share of jobs Changed: Testsuite Review: CR MD-2005-04-21-0: Bugfix: - fixed windows execd loadsensor path - README Changed: arch_variables, inst_common.sh, README-Autoinst JG-2005-04-21-0: Bugfix: Fixed memory leak in scheduler. Changed: sge_schedd Review: SG CR-2005-04-21-0: Testsuite: - added new test: functional/consumable_resources - handle_vi_edit(): o function did not export env() to open_remote_spawn_process() o added TODO comment o fixed some missleading output - set/get_complex(): o moved old (unverified) code to sge_procedures.53.tcl o modified functions for 60 version - get_qstat_j_info(): o schedd info string did not contain job resource schedd message because this message contains a ":" which evoke a parse error Issue: 1212 (not closed because there are still some enhancments to implement) RD-2005-04-20-0: CVS Created V60u4rc1_TAG --------------------------------------------------------- MD-2005-04-20-1: Bugfix: changed README MD-2005-04-20-0: Bugfix: - inst_sge -bup with rpc server destroy database - README Autoinst, README Upgrade Bugtraq: 6257389 Issue: 1576 Changed: inst_common.sh Review: RD MD-2005-04-19-0: Enhancm.: - Windows enhancement - testsuite (windows enhancement) - autoinstall error message enhancement - clean up - autoinstall readme Review: RD RD-2005-04-19-1: Bugfix: qconf -purge segfault on tru64 Bugtraq: 6252522 Issue: 1548 Review: AH CR-2005-04-19-1: Testsuite: o fixed cleanup function for issue test 1096 when running in CSP mode (second fix) RD-2005-04-19-0: Bugfix: fixed broken manual bdb rpc installation Review: CR JG-2005-04-19-0: Testsuite: Fixed testsuite for cluster with hostaliasing. Fixed test bugs/issuezilla/403 (now correctly detects reopened issue 403). Cleanup qsub test. CR-2005-04-19-0: Testsuite: o fixed cleanup function for issue test 1096 when running in CSP mode o Enhancem.: Added new function: copy_certificates (copy CSP (SSL) certificates to a host) CR-2005-04-18-4: Testsuite: fixed test system_tests/commlib_issues when module test binary can't malloc() enough memory JG-2005-04-18-0: Testsuite: fixed test functional/tight_integration CR-2005-04-18-3: Testsuite: new menue parameter: "check spool directory content" AH-2005-04-18-0: Bugfix: cqueues/all.q trashed after qmaster shutdown with 1362 hosts Changed: qmaster Review: EB, JG Bugtraq: 6256530 Issue: 1572 RD-2005-04-18-1: Bugfix: autoinstall related bugs fixed bdb rpc auto installation and uninstallation, hostname resolving fixes, some cleanups Changed: install scripts Bugtraq: 6253278, 6253291, 6253192 Issue: 1573, 1574, 1575 Review: MD CR-2005-04-18-2: Cleanup: removed unused message macro definition CR-2005-04-18-1: Cleanup: o qmake cleanup Testsuite: o removed warning messages for unexpected qmake output Review: JG CR-2005-04-18-0: Bugfix: Fixed broken build for commlib issues module test SG-2005-04-15-3: Bugfix: fixed SG-2005-04-15-2 Changed: qmaster Review: pending CR-2005-04-15-3: Bugfix: qmake in dynamic allocation mode core dump Changed: qmake Bugtraq: 6255902 Issue: 1568 Review: JG CR-2005-04-15-2: Testsuite: - thight integration test blocked killing of slave/master task SG-2005-04-15-2: Bugfix: the usage in projects is never spooled while the qmaster is alive. Changed: qmaster Issue: 1567 Bugtraw: 6255850 Review: JG DT-2005-04-15-0: Bugfix: o Fixed qsub issue with parsing large scripts Changed: parse_job_cull Review: CR SG-2005-04-15-1: Bugfix: execd does sends empty job report for a pe slave task Changed: execd Issue: 1563 Bugtraq: 6255336 Review: JG SG-2005-04-15-0: Bugfix: job in error state breaks qstat -f -xml Changed: qstat Issue: 1566 Bugtraq: 6255804 Review: CR CR-2005-04-15-1: Enhancem.: - improved commlib error message output for connection errors. (Now commlib calls strerror() to report system error message) Changed: commlib (all binaries) Review: JG CR-2005-04-15-0: Bugfix: - fixed memory leak in commlib (SSL Framework) - fixed memory leak in qstat Changed: commlib (all binaries), qstat Review: JG MD-2005-04-14-0: Bugfix: - inst_sge only deletes common - auto_inst uses ADMIN_HOST_LIST variable only at qmaster installation time Changed: inst_sge Issue: 1356, 1565 Bugtraq: 6239655, 6253138 Review: RD SG-2005-04-14-0: Bugfix: qmaster does not store sharetree usage on shutdown Changed: qmaster Issue: 1562 Bugtraq: 6255329 Review: JG CR-2005-04-14-2: Testsuite: o modified remote_procedures (can block shell) shift login calls/tests into testsuite script files to have always a /bin/sh DT-2005-04-14-1: Testsuite: o Checked in new test jar for DRMAA-J RD-2005-04-14-1: Cleanup: minor changes Review: CR RD-2005-04-14-0: Bugfix: auto_install fixes -um now uses configuration file BDB RPC NFS spooling recognized Changed: install scripts Bugtraq: 6253313, 6253219 Issue: 1559, 1560 Review: MD DT-2005-04-14-0: Bugfix: o Updated version string for DRMAA-J Changed: DRMAA-J CR-2005-04-14-1: Testsuite: o midified commlib issues module test to be more error tolerant (when some hosts have to less memory (only 256 MB) this test fails, because they're starting to swap. CR-2005-04-14-0: Testsuite: o modified remote_procedures (can block shell) (increased timeout) DT-2005-04-13-1: Bugfix: o Fixed minor DRMAA-J issue Doc: o Checked in current JavaDocs Changed: DRMAA, doc DT-2005-04-13-0: Bugfix: o Fixed DRMAA-J testsuite issues Testsuite: o Fixed minor DRMAA-J issues Changed: DRMAA, Testsuite Review: AH JG-2005-04-13-1: Doc: Fixed typo in man page sge_conf.5 CR-2005-04-13-0: Testsuite: o modified remote_procedures (can block shell) o modified timeout setting for submit_job JG-2005-04-13-0: Testsuite: Check bugs/issuezilla/1473 failed with host aliasing. Some cleanup in system_tests/clients/qsub. RC-2005-04-13-1: Enhancem.: Failed to compile on Cray unicos Changed: - sge_getloadavg.c (iolist instead of iolistr) RC-2005-04-13-0: Enhancem.: - arch should use uname -p to determine NetBSD platform - identify DragonFly as FreeBSD (Joshua Coombs) Review : BD, NJ CR-2005-04-12-2: Testsuite: enhanced usage test error output SG-2005-04-12-0: Bugfix: fixed build for lx26-x86 CR-2005-04-12-1: Testsuite: - cleanup function was called, even when test was not running - qsub test: retry to submit with qsub -now yes option due to bugtraq bug 4938859 Each retry reports a warning message, test fails for more than 3 retries. CR-2005-04-12-0: Testsuite: Modified tight integration test to report only warning when there are messages logged into the execd messages file. RD-2005-04-12-0: Bugfix: Fixed broken build DT-2005-04-11-0: Bugfix: o Removed static array from option file parsing Issue: 1551 BugTraq: 6252465 Changed: parse_job_cull Review: SG SG-2005-04-11-0: Bugfix: failed array tasks are rescheduled only one by one Review: AH Bugtraq: 6253266 Issue: 1554 Changed: scheduler JG-2005-04-11-0: Testsuite: Fixed test system_tests/qmaster/size for sol-amd64 AH-2005-04-11-0: Bugfix: qstat -f -pe make breaks Review: EB Bugtraq: 6253093 Issue: 1552 Changed: qstat JG-2005-04-11-0: Bugfix: Missing success message with qconf -Aprj Changed: qconf Review: AD Issue: 1549 Bugtraq: 6252524 MD-2005-04-11-1: Bugfix: install_qmaster picks up commented out service sge_qmaster Changed: inst_common.sh Bugtraq: 6251178 Issue: 1540 Review: HP MD-2005-04-11-0: Bugfix: berkeleydb server shutdown script failes Changed: sgebdb rc-template Bugtraq: 6251175 Issue: 1545 Review: HP RD-2005-04-11-0: Bugfix: qconf -purge core dump with last hostlist attribute Changed: qconf Bugtraq: 6252522 Review: RH RC-2005-04-11-1: Enhancem.: Fixed some spelling mistakes Changed: rmon.html RC-2005-04-11-0: Bugfix: arch.dist is broken on *BSD Review: AH Changed: arch.dist RD-2005-04-07-0: Bugfix: commlib deadlock with host aliasing Review: CR Bugtraq: 6251943 Changed: commlib CR-2005-04-06-8: Testsuite: fixed testsuite test commlib/issues and commlib test_issues module test CR-2005-04-06-7: Testsuite: added missing file for qsub tests (-sync) CR-2005-04-06-6: Testsuite: qdel test: delete all jobs before starting next sub test CR-2005-04-06-5: Testsuite: test for issue 1096 disabled for CSP systems CR-2005-04-06-4: Testsuite: fixed job_environment test AH-2005-04-06-0: Testsuite: fixed module test test_sge_qmaster_timed_event SG-2005-04-06-0: Bugfix: reserved jobs prevent other jobs from starting Review: AH Issue: 1543 Bugtraq: 6251172 Changed: scheduler CR-2005-04-06-3: Testsuite: fixed testsuite test bugs/issuezilla/1096 for BDB RPC spooling CR-2005-04-06-2: Testsuite: fixed testsuite test bugs/issuezilla/1334 CR-2005-04-06-1: Testsuite: rlogin timeout problem in open_remote_spawn_process() CR-2005-04-06-0: Testsuite: install_core_system test: - splited mask check into two start_remote_prog calls - add submit only hosts to submit host list at execd installation - fixed bug in inst_sge version check handling Review: MD EB-2005-04-05-0: Enhancem.: added loadsensor script for windows systems Review: HP EB-2005-04-03-0: Enhancem.: added functionality to windows loadsensor Review: HP HP-2005-04-04-1: Cleanup: removed double WIN32 block, added TMPDIR for faster compiling with gcc compilers Changed: aimk.site Review: EB HP-2005-04-04-0: Bugfix: scripts/distinst now copies ASCMAN/man/manX to man/catX for Interix man. Bugtraq: 6239465 Changed: scripts/distinst Review: EB JG-2005-04-04-0: Testsuite: Fixed check bugs/issuezilla/1489 MD-2005-04-01-0: Testsuite: BDB RPC Server installation support Changed: testsuite Review: CR MD-2005-03-31-0: Bugfix: - upgrade failed at sharetree - inst_sge -x fails with fqdn - autoinst does not fully consider ignore_fqdn - load.sh returns wrong load values, if uptime higher than 1 hour Issue: 1494, 1478, 1119, 1535 Bugtraq: 6233300, 6239640, 5071527, 5086193 Changed: inst_sge, install_modules Review: CR CR-2005-03-31-1: Testsuite: increased timeout waiting after qconf -km to shutdown qmaster with kill signal CR-2005-03-31-0: Cleanup: o Using SSL_set_quiet_shutdown() to shutdown a broken SSL connection. (GE CSP mode) o Added hostname into SSL accept error messages Review: MD Changed: commlib DT-2005-03-31-0: Bugfix: o Fixed qsub -sync return codes o Added test suite tests Issue: 1433 BugTraq: 6221231 Changed: qsub, man Review: SG JG-2005-03-31-0: Bugfix: Fixed qmaster profiling: - profiling was not activated at qmaster startup - cpu stime and utime are not per thread values Issue: 1486, 1487 BugTraq: 6239642, 6239660 Changed: sge_qmaster, man Review: SG AH-2005-03-30-0: Doc: Fixes with man pages: o mismatch of man page for host_aliases o bad example in man page for complex o "qconf -Aconf " error o sge_shadowd(8) man page should be improved Issue: 932, 925, 890, 497 BugTraq: 4820420 Changed: man Review: CR CR-2005-03-30-0: Bugfix: o Increased GDI version (new ssl layer) o After shadowd takes over there is a long delay before execd connects to new qmaster Issue: 1533 Bugtraq: 6222930 Changed: gdi lib, execd Review: AH RH-2005-03-30-0: Enhancem.: com_sun_grid_util_SGEUtil_getPID method added to libdrmaa.so Review: DT CR-2005-03-29-0: Cleanup: cleanup of ssl message logging Review: SG CR-2005-03-29-0: Bugfix: sequence nr of execd load reports corrupted Bugtraq: 6247239 Issue: 1450 Changed: execd Review: SG DT-2005-03-29-2: Bugfix: o Fixed broken build on MacOS X & HP-UX Changed: profiling DT-2005-03-29-1: Bugfix: o Fixed broken build on Linux Review: CR Changed: profiling DT-2005-03-29-0: Bugfix: o Profiling is now thread-safe and DRMAA-safe o Fixed MT bug in drmaa_exit() o Profiling & DRMAA cleanup o Fixed qsub -b n -cwd Issue: 1471, 1529 BugTraq: 6242165, 6247238 Changed: DRMAA, profiling, read_defaults Review: SG, MD SG-2005-03-29-0: Bugfix: qstat -explain E does not print queue errors correctly Bugtraq: 6247211 Issue: 1525 Changed: qmaster Review: DT CR-2005-03-25-0: Bugfix: Fixed broken sge_ca (CSP mode) script AA-2005-03-23-0: Bugfix: Certificate renewal in CSP mode broken Bugtraq: 6232120, 6245486 Issue: 1515, 1395 Changed: sge_ca, sge_ssl.cnf, sge_ssl_template.cnf, renew_all_certs.csh Review: CR EB-2005-03-24-0: Bugfix: qhost -h does not show selected host Changed: qhost Issue: 1524 Bugtraq: 6245487 Review: SG JG-2005-03-24-0: Testsuite: Testsuite didn't compile / install binaries for bdb_server and submit_only_hosts. Issue: 1523 Fixed checks functional/usage and performance/throughput Cleanup. RC-2005-03-24-3: Bugfix: allow installation of binaries compiled on Linux 2.6 kernels Review: AS, JM Issue: 1310, 1367 Changed: arch.dist RC-2005-03-24-2: Enhancem.: Failed to compile on Cray unicos Changed: - pdc.c (include file is now sge_unistd.h) - sge_c_gdi.c, configuration_qmaster.c (conflict in function prototype: "static" missing) RC-2005-03-24-1: Enhancem.: moved function declarations into sge_uidgid.h Changed : get*_r routines for BSD (and other OSes with them missing) RC-2005-03-24-0: Bugfix: - fixed a typo (should be qtcsh instead of qtsch) - qtcsh failed to compile on PPC Linux - restored Changelog Changed: qtcsh CR-2005-03-24-0: Bugfix: forgot to initalize connection response state at connection setup. Changed: commlib Review: RH AH-2005-03-23-0: Bugfix: a series of matching soft queue requests gets not counted separately Review: SG Issue: 1520 Bugtraq: 6244865 Changed: sge_schedd JG-2005-03-23-0: Bugfix: qdel failed to delete qrsh (login) job on a Solaris box when Secure Shell is used Review: AH Issue: 1519 Bugtraq: 6215730 Changed: sge_shepherd SG-2005-03-23-0: Bugfix: scheduler does not get all objects on a qmaster or scheduler startup Review: DT Issue: 1518 Bugtraq: 6244808 Changed: qmaster CR-2005-03-23-0: Bugfix: error message from execd about endpoint is not unique Issue: 1504 Bugtraq: 6234371 Changed: commlib, all binaries Review: MD RH-2005-03-23-0: Bugfix: misleading qstat -j message when the scheduler is not running Issue: 1435 Bugtraq: 6244229 Changed: qmaster Review: SG RH-2005-03-22-1: Bugfix: qsub -b y must fail if no command is specified Cleanup: shepherd_trace calls ifdef'ed Issue: 1365 Bugtraq: 6244215 Changed: qsub, shepherd Review: AH DT-2005-03-22-0: Bugfix: o Fixed broken build Changed: Shepherd JG-2005-03-22-0: Testsuite: Added support for solaris zones. Issue: 1503 RH-2005-03-22-0: Bugfix: starter_method is ignored with binary jobs that are started without a shell Issue: 1334 Bugtraq: 6194719 Changed: shepherd Review: AA AA-2005-03-21-1: Bugfix: - loading of shared openssl libraries on Mac and AIX - clean rule: remove generated c file Issue: 1354 Review: CR Changed: all bins AA-2005-03-21-0: Bugfix: - make qtcsh compile Changed: qtcsh Review: CR CR-2005-03-21-2: Enhancem.: - (gridengine CSP mode) added error message for invalid certificates CR-2005-03-21-1: Bugfix: - removed unused message macros, fixed not unique macro message id CR-2005-03-21-0: Testsuite: - added new command line argument: compile_clean (do a aimk clean even when no_update option is set) - aimk clean did not use aimk compile options from testsuite settings file - install_core_system test (CSP mode): wait for qstat -f working after certificates has been copied to remote hosts RH-2005-03-21-0: Bugfix: termination script may not be ignored, when job submited with -notify Issue: 1489 Bugtraq: 6241487 Changed: sheperd Review: DT RC-2005-03-21-1: Bugfix: malloc did not check for NULL Changed: execd (Solaris) RC-2005-03-21-0: Bugfix: failed to build on IA64 and Alpha Linux Changed: 3rdparty/qtcsh/sh.h CR-2005-03-18-2: Bugfix: fixed CSP problems on hp11/aix43/aix51 Review: AA CR-2005-03-18-1: Bugfix: additonal changes for CR-2005-03-17-0 Changed: qmaster hard/soft limit limit setting Issue: 1502 Bugtraq: 6239569 Review: AA CR-2005-03-18-0: Testsuite: - new command line parameter: no_local_spool: don't use local spool directories - fixed bug in install_core_system MD-2005-03-17-4: Enhancm: Request for start-up script additions / Use LSB installation procedure for rc scripts when available - minor bugfixes testsuitetest IZ: 1384 Changed: rc templates, inst_common.sh RFE IZ: 1424 Bugtraq: 6221850 Review: HP MD-2005-03-17-3: Bugfix: install_execd does not pick up $SGE_CELL Changed: inst_execd.sh Issue: 1493 Bugtraq: 6197109 Review: HP MD-2005-03-17-2: Bugfix: install script should create execd spooldir Changed: inst_execd.sh Issue: 1492 Bugtraq: 6235845 Review: HP MD-2005-03-17-1: Bugfix: adminuser is not considered in autoinstall! Changed: auto_inst template, inst_common.sh Issue: 1243 Bugtraq: 6239504 Review: HP MD-2005-03-17-0: Bugfix: execd installation fails, if local spool dir is not entered Changed: inst_execd.sh Issue: 1384 Bugtraq: 6203977 Review: HP CR-2005-03-17-1: Bugfix: change for CR-2005-03-17-0 Changed: qmaster hard/soft limit limit setting Issue: 1502 Bugtraq: 6239569 HP-2005-03-17-0: Bugfix: Replaced direct compare of UID to 0 with function call to sge_is_start_user_superuser(), removed sge_is_start_user_root() from library because it did the same / Put error message in Debug output in error handling branch. Changed: libs/uti/sge_uidgid.c, libs/uti/sge_io.c Review: RD CR-2005-03-17-0: Bugfix: qmaster does not accept new connections if number of execd's exceed FD_SETSIZE Issue: 1502 Bugtraq: 6239569 Review: AA Changed: aimk commlib qmaster qmon SG-2005-03-17-1: Bugfix: jobs which request a PE and a consumable resources which set to infinity in global host configuration are not scheduled Issue: 1447 Bugtraq: 6242057 Review: DT Changed: scheduler SG-2005-03-17-0: Bugfix: Consumable request may not be 0 if PE requested / New Min-Max entry in complex definition Issue: 1379 Bugtraq: 6242055 Review: DT Changed: scheduler RH-2005-03-17-0: Bugfix: releasing user hold state through qrls may not require manager privileges Issue: 1434 Bugtraq: 6221244 Changed: qalter, qlrs, qhold, qmaster, testsuite (qalter test) Review: EB SG-2005-03-16-3: Bugfix: qstat -F dies in case of a infinit integer setting for a complex attribute Issue: 1508 Bugtraq: 6241544 Review: JG Changed: qstat, qmaster, scheduler, ... RD-2005-03-16-0: Bugfix: Spooledit fails during database upgrade RD-2005-03-11-0 had problems with string which includes quotations. Issue: 1425 Bugtraq: 6239394 Review: JG Changed: spooledit SG-2005-03-16-2: Bugfix: error message "no execd known on host" Issue: 1431 Bugtraq: 6241430 Review: JG Changed: qmaster SG-2005-03-16-1: Bugfix: Conflicting requirements should have the same meaning with qstat and qsub Issue: 1462 Bugtraq: 6241401 Review: EB Changed: qstat, qhost, qacct, qmon SG-2005-03-16-0: Bugfix: qselect uses sge_qstat file Issue: 1505 Bugtraq: 6229277 Changed: qstat Review: EB JG-2005-03-15-2: Testsuite: Added zones parameter to the testsuite host config. Issue: 1503 Fixed check bugs/issuezilla/1473/check.60.exp. EB-2005-03-15-0: Bugfix: qstat -s hu shows pending jobs only Issue: 1439 Bugtraq: 6240739 Changed: qstat Review: SG JG-2005-03-15-1: Bugfix: qconf -Mc only outputs one line, even if multiple complex attributes are added/deleted/modified. Issue: 1506 Bugtraq: 6178843 Changed: qconf Review: SG SG-2005-03-15-2: Bugfix: qstat -U aborts Issue: 1469 Bugtraq: 6241376 Changed: qstat Review: EB SG-2005-03-15-1: Bugfix: reprioritize parameter is NOT documented Issue: 1501 Bugtraq: 6229603 Changed: man5/sge_conf.5 Review: DT SG-2005-03-15-0: Bugfix: Reservation of wrong hosts Issue: 1484 Bugtraq: 6241378 Changed: scheduler Review: EB JG-2005-03-15-0: Bugfix: Huge cpu and memory overhead when adding, modifying and deleting complex attributes. Issue: 1473 Bugtraq: 6222237 Changed: qmaster Review: SG RD-2005-03-14-0: Bugfix: amendmand for RD-2005-03-11-0 Review: AA RC-2005-03-13-2: Enhancem.: NetBSD port (forward porting from SGE 5.3) RC-2005-03-13-1: Bugfix: fixed build (uninitialized variable) Review: Reuti RC-2005-03-13-0: Enhancem.: created target depend for qmake Review: Reuti SD-2005-03-12-0: Enhancem.: Added support for tracking pthreads to the Linux PDC (portable data collector). RC-2005-03-12-1: Enhancem.: PPC Linux port RC-2005-03-12-0: Enhancem.: ignore warnings when calling configure for qmake/qtcsh. (only affects non-standard platforms with no makefile in cvs) RD-2005-03-11-1: Bugfix: fixed build for lx24-amd64 & lx24-x86 Review: CR,RH RD-2005-03-11-0: Bugfix: Spooledit fails during database upgrade Issue: 1425 Bugtraq: 6239394 Review: JG RH-2005-03-11-1: Enhancem.: checkprog accepts now the options -ppid. The parent process will be printed to stdout Changed: checkprog Review: AS RD-2005-03-10-1: Bugfix Problems with shadowd install Issue: 1347 Bugtraq: 6197730 Review: MD RD-2005-03-10-0: Bugfix qstat -f -q all.q@HOSTNAME does not resolve hostname Issue: 1220 Bugtraq: 5085004 Review: CR CR-2005-03-10-0: Bugfix: - fixed thread signal behaviour for application threads calling cl_commlib_receive_message() at the same time. - fixed qping failure when no sge environment is sourced Review: MD JG-2005-03-09-0: Enhancem.: More fine grained profiling information in qmaster. Profiling for the spooling of jobs scripts. Review: MD JG-2005-03-08-0: Testsuite: Make all tests use the setup/cleanup framework. Issue: 888 JG-2005-03-08-0: Bugfix: Fixed broken build. DT-2005-03-04-0: Bugfix: o Cleaned up JAPI locking o Cleaned up JAPI session handling o Cleaned up DRMAA-J JNI code o Fixed security bugs in DRMAA-J o Fixed segfault in profiling o Fixed various DRMAA-J bugs Changed: DRMAA-J, JAPI, profiling Issue: 1470, 1472, 1422, 1076, 1479 BugTraq 6236475, 6236476, 6236472, 6236469, 6236479 Review: SG, MD SG-2005-03-04-0: Enhancem.: - added async-gdi handling - enhanced the job class scheduler - cleanup in the order handling Changed: qmaster, scheduler Review: JG (async-gdi) DT (cleanup, job-classes) RD-2005-03-04-0: Enhancm.: BDB install on NFSv4 share fstype now reports nfs4 on solaris installation recognize nfs4 as valid Issue: 1448 Bugtraq: 6236261 Review: AA MD-2005-03-04-0: Bugfix: - Port free/used check returns a wrong result in some cases (1385) - execd uninstall doesn't remove all objects (1208) - backup/restore for classic and rpc server spooling not supported! (1244) - restore procedure does not really ensure qmaster is down (1242) Issue: 1208, 1242, 1244, 1385 Changed: inst_sge, install_modules, getservbyname Bugtraq: 6203984, 6231589, 6236139, 6236136 Review: RD JG-2005-03-03-0: Testsuite: Make more tests use the setup/cleanup framework. Issue: 888 SG-2005-03-03-1: Cleanup: minor code changes. Review: DT changed: all RD-2005-03-03-0: Bugfix: purge don't clean hostlist attribute Review: MD SG-2005-03-03-0: Bugfix: fixed locking issues in the commlib Review: AA Changed: all SG-2005-03-02-1: Cleanup: Changed MAXHOSTLEN to CL_MAXHOSTLEN due to a collision between different defines under linux. Review: AA Changed: all JG-2005-03-02-0: Testsuite: Provide means to configure explicit setup and cleanup functions both for whole checks and per check level. Changed a number of checks to use this feature, other checks still have to be changed. Issue: 888 RD-2005-03-02-0: Cleanup: compile with gcc-3.3 Issue: 558 Review: SG SG-2005-03-02-0: Bugfix: fixed build for lx24-amd Review: RD RD-2005-03-01-1: Enhancem.: -purge switch added purge host specific cluster queue profiles Issue: 1453 Bugtraq: 6234836 Review: MD JG-2005-02-28-1: Testsuite: Adding high numbers of hosts during qmaster installation could break communication between expect and install script. JG-2005-02-28-0: Testsuite: Added support for architecture sol-amd64. RD-2005-02-25-0: Bugfix: Problems with BDB RPC server are hard to diagnose Changed: bdb_rc_template Issue: 1418 Bugtraq: 6218379 Review: AH SG-2005-02-25-0: Bugfix: global scheduler messages are reported multiple times Changed: scheduler Issue: 1468 Bugtraq: 6233162 Review: JG JG-2005-02-25-0: Bugfix: Adding/modifying a high number of complex entries took a long time, qconf could timeout. First part of fix: Reduce the number of situations, where a recomputation of consumables is required. Changed: qmaster Issue: 1473 Bugtraq: 6222237 Review: SG JM-2005-02-23-1: Enhancem.: Added support to handle setup of startup scripts in a Debian-specific way (update-rc.d). Changed: arch_variables, inst_common.sh Reviewed: RH,AH SG-2005-02-23-0: Bugfix: load formula is not working for pe jobs Changed: scheduler Bugtraq: 6232074 Issue: 1454 Review: AH JG-2005-02-22-0: Enhancem.: Improved performance of search for duplicate centries. Does not yet fix the issues listed below, but provides some improvement. Issue: 1473 Bugtraq: 6222237 Review: SG SG-2005-02-21-0: Bugfix: - deadlock in the qmaster with qconf -ks - reworked potential deadlocks between the two GDI threads, and/or a GDI and the Event Master thread. Changed: qmaster Bugtraq: 6231366 Issue: 1458 Review: DT SG-2005-02-18-0: Bugfix: execd logs error mesage, when a tight pe job in "t" state is deleted Changed: qmaster Bugtraq: 6230846 Issue: 1465 Review: AA AH-2005-02-15-0: Bugfix: qconf error diagnosis broken Changed: qconf Bugtraq: 6186597 Review: HP SG-2005-02-15-0: Bugfix: An array pe job can set queues into error state Changed: execd Issue: 1463 Bugtraq: 6229373 Review: DT SG-2005-02-15-0: Bugfix: a parallel arrayjob can kill the qmaster Changed: qmaster Issue: 1441 Bugtraq: 6229253 Review: AH AH-2005-02-14-0: Bugfix: Execd messages file contains incorrectly-formatted lines Changed: execd Bugtraq: 6228350 Review: SG HP-2005-02-14-0: Bugfix: Changed behaiour of uidgid for Windows, so that now 0 is returned for an Administrator-UID. Changed: source/utilbin/uidgid.c SG-2005-02-14-1: Bugfix: some changes to build the source on sol10-amd64 Changed: qmon RD-2005-02-14-0: Bugfix: add warnings for execd_spool dir change Changed: qmaster, execd Issue: 103 Bugtraq: 6219999 Review: AH SG-2005-02-14-0: Bugfix: Long delay when submitting large pe jobs Changed: qmaster Issue: 1461 Bagtraq: 6228786 Review: DT HP-2005-02-14-0: Bugfix: Set arch_variables for Windows Changed: dist/util/arch_variables AA-2005-02-11-0: Bugfix: amd64 bit qmon, what a stupid problem a wrong define in aimk (MOTIF12_HEADERS) was the reason for the problems under amd64 Latest build of libXm.so on dain installed Changed: aimk, qmon SG-2005-02-07-4: Bugfix: sharetree has a usage leak Changed: qmaster Issue: 1449 Bugtraq: 6225570 Review: DT SG-2005-02-07-3: Bugfix: some changes to build the source on sol10-amd64 missed some files in (SG-2005-02-07-0) Changed: all SG-2005-02-07-2: Bugfix: suspend_internal is ignored when enabling jobs due to suspend_thresholds change Changed: qmaster Issue: 1436 Bugtraq: 6226085 Review: DT SG-2005-02-07-1: Bugfix: init variables in the execd Changed: execd Review: DT SG-2005-02-07-0: Bugfix: some changes to build the source on sol10-amd64 Changed: all SG-2005-02-04-0: Bugfix: - fixed build - fixed wrong memory access Changed: all Review: AH CR-2005-02-04-1: Enhancem.: added warning messages when load reports get lost. Issue: 1450 Review: SG CR-2005-02-04-0: Enhancem.: - use qping as rmon debug client (part 1) - qping rmon client support can be enabled by setting -DDEBUG_CLIENT_SUPPORT at compile time - qping now can send debug instructions to the connected endpoint Review: MD CR-2005-02-03-0: Bugfix: fixed broken build for hp11 Review: AA RD-2005-02-03-0: Enhancem.: - code cleanup kvm dependency and code removed Review: MD EB-2005-02-03-0 Enhancem.: set rsh_daemon=/usr/sbin/in.rshd for Interix CR-2005-02-02-0: Enhancm.: reimplemenation and integration of former csp security into commlib layer Changed: gdi, commlib Review: AA RD-2005-02-02-0: Bugfix: switched use of /dev/kmem to kstat Issue: 1403 Bugtraq: 6218430 Changed: loadcheck, execd Review: JG, CR MD-2005-01-31-0: Bugfix: distinst assumes certain files always exist Issue: 1361 Review: AS AH-2005-01-28-0: Bugfix: Arguments to binaries sent to qsub are given to invoking shell too Issue: 1442 Review: GU SG-2005-01-28-0: Bugfix: - added a check for a valid scheduler config. Review: pending Changed: scheduler, qconf Issue: 1438 Bugtraq: 6222811 SG-2005-01-28-0: Bugfix: - fixed qstat filtering. SG-2005-01-21-0 did not fix all issues Review: EB Changed: qstat Issue: 1420 Bugtraq: 6218877 AS-2004-11-26-0: CVS Created V60_s10amd64_beta_TAG --------------------------------------------------------- CR-2005-01-26-0: Enhancem.: - qping -dump now supports application debug messages Review: MD CR-2005-01-25-1: Enhancem.: - code cleanup CR-2005-01-25-0: Bugfix: - fixed qping connect problem checked in with CR-2005-01-21-0 Enhancem.: - qping to a running service (e.g. telnet) did not cancel when pressing CTRL + C. Review: SG SG-2005-01-25-0: Bugfix: the scheduler crashed in case of a restart and a running MPI job. Changed: scheduler Review: EB Issue: 1416 Bugtraq: 6221167 AS-2005-01-25-1: Bugfix: set file permissions in destination directory" Changed: scripts/distinst Review: MD AA-2005-01-24-0: Bugfix: labels of AFS section shwon twice Changed: qmon CR-2005-01-21-0: Enhancem.: - finished commlib SSL framework support ( not yet integrated in gdi ) - code cleanup Review: AA SG-2005-01-21-1: Bugfix: - qmaster crashed in case of some calendar configs. Review: AH Changed: qmaster Issue: 1427 Bugtraq: 6220060 SG-2005-01-21-0: Bugfix: - fixed qstat -t - fixed a minor uninit variables in the execd Review: DT Changed: qstat, execd Issue: 1420 Bugtraq: 6218877 RD-2005-01-19-0: Bugfix: CULL examples fixed Review: AH EB-2005-01-19-0: Enhancem.: qtcsh port for WIN32_X86 Changed: qtcsh DT-2005-01-19-0: Testuite: o Fixed drmaa-j test DT-2005-01-18-0: Cleanup: o Made JAPI Makefile more readable o Made class and jar file compilation cleaner Bugfix: o Added new test target for Ant for testing binaries Changed: DRMAA RD-2005-01-17-0: Bugfix: make qmon compile on sol-amd64 Changed: qmon HP-2005-01-17-0: Bugfix: qmake Makefile for WIN32_X86 now with relative pathes Changed qmake DT-2005-01-17-0: Bugfix: o JAPI makefile didn't delete drmaa.o on clean Changed: DRMAA CR-2005-01-17-0: Testsuite: cleanup for issue check 126 DT-2005-01-16-0: Enhancem.: o aimk now can build javadocs with or without protected fields and methods. Bugfix: o Changed Session.getVersion() to return the binding spec version. Changed: DRMAA, aimk HP-2005-01-15-0: Enhancem.: Changed aimk in WIN32_X86 section Changed: aimk CR-2005-01-14-0: Testsuite: Modified test for issue #126 to report more error information into the testsuite error mail message. (using qping -dump) HP-2005-01-12-0: Enhancem.: Changed aimk in WIN32_X86 section Changed: aimk CR-2005-01-13-0: Bugfix: - qping -dump hex buffer converting error from CR-2004-12-01-0 (only concerns maintrunk) - code cleanup Changed: commlib, qping, execd, qmaster Review: HP AH-2005-01-12-0: Doc: Errors with load_avg in sched_conf(5) sched_conf(5) wrongly mentioned load_avg rather than np_load_avg and 100 rather than 1.0 for a job that utilizes a full cpu Changed: man Review: AS Issue: 1417 Bugtraq: 6216694 CR-2005-01-12-0: Enhancem.: commlib read/write errors closing connection messages in the daemon log files will now also contain the name of the connection where the read/write error appeared. Changed: commlib Review: AA CR-2005-01-11-2: Testsuite: added testsuite code for parsing qping -dump output into tcl_files/remote_procedures.tcl CR-2005-01-11-1: Testsuite: modified issue test #141 for detection of issue #1414 CR-2005-01-11-0: Bugfix: pending job task deletion may not work Changed: qmaster Issue: 1414 Bugtraq: 6216020 Review: SG EB-2005-01-11-0: Enhancem.: Merged missing code from V60_win_alpha_BRANCH. HP-2005-01-10-2: Enhancem.: Optimized aimk section for WIN32_X86 Changed: aimk HP-2005-01-10-1: Added: qmake configure files for Interix HP-2005-01-10-0: Enhancem.: Changed name of function in libwingrid Changed: sge_load_sensor.c EB-2005-01-10-0: Enhancem.: Changed distinst for Interix installation Review: HP CR-2005-01-07-0: Bugfix: - master task will close connection to execd after job delivery Changed: qsh, gdilib Issue: 1413 Bugtraq: 6215580 Review: SG CR-2005-01-07-0: Testsuite: enhanced tight integration check to report errors for issue #1413 DT-2005-01-05-2: Enhancem.: o aimk now checks the Java version Changed: aimk.site DT-2005-01-05-1: Bugfix: o Fixed DRMAA-J PartialTimestamp bugs Docs: o Added section to DRMAA-J howto Changed: DRMAA, www DT-2005-01-05-0: Bugfix: o Fixed minor DRMAA-J bugs o Removed JGDI flags from aimk o Reduced MAX_STRING_SIZE to 2048 o Added DRMAA-J testsuite test o Made symlink to JNI lib on MacOS X Changed: DRMAA, aimk, basis_types, job, stdlib, testsuite, distinst Issue: 538 Review: SG, AS HP-2005-01-04-1: Bugfix: Fixed bug in aimk in code for Interix and Win32 Changed: aimk HP-2005-01-04-0: Bugfix: Improved "root" or "Administrator" detection in Interix. Changed: libuti, shepherd CR-2005-01-03-0: Testsuite: - added binary test module for issue #1389 and issue #1400 - added testsuite test for test module JG-2005-01-03-1: Bugfix: Fixed broken build. JG-2005-01-03-0: Bugfix: Fixed aimk -no-mk. DT-2005-01-03-0: Docs o Updated DRMAA-J howto and README o Removed javadoc.clean from clean target Bugfix: o Fixed missing qtask args in qtcsh - Contributed by Ron Chen o Fixed colon parsing in working directory in DRMAA - Contributed by Ron Chen Changed: DRMAA, www, qtcsh Review: DT Issue: 1407, 1397 DT-2004-12-31-1: Bugfix: o Really backed out conflicting changes o Resolved conflict between ARCo and DRMAA Changed: aimk.site, DRMAA Makefiles DT-2004-12-31-0: Bugfix: o Backed out some changes from DT-2004-12-29-0 which conflicted with aimk -j 2 Changed: DRMAA Makefiles DT-2004-12-30-2: Docs: o Updated JavaDocs Changed: javadocs, DRMAA DT-2004-12-30-1: Bugfix: o Fixed broken Java build Changed: aimk, DRMAA Review: pending DT-2004-12-30-0: Bugfix: o Fixed minor bug in aimk o Fixed minor bug in distinst o Backed out part of DT-2004-12-29-0 o un-DOS'ed JobTemplateImplTest.java o Fixed typos in README.aimk Changed: DRMAA, aimk, zerodepend, distinst, README.aimk Review: pending DT-2004-12-29-0: Bugfix: o Restructured DRMAA build proceedure o Fixed minor bug in findcheckin script Changed: DRMAA, aimk, distinst, zerodepend Issue: 1405 Review: pending DT-2004-12-28-0: Bugfix: o Added missing files from previous checkins o Fixed bugs in DRMAA-J test suite o Fixed packaging for DrmaaExample and TestDrmaa o Fixed DRMAA-J cloning issue - Still need to test new clone() methods Changed: DRMAA Issue: 1404 CR-2004-12-23-0: Bugfix: qmaster running out of file descriptors - Part II of CR-2004-12-03-2 (added additional file desriptor soft limit check at qmaster startup) Issue: 1400 Bugtraq: 6211309 Changed: qmaster, commlib Review: MD JG-2004-12-23-0: Testsuite: Do not check directory permissions and messages files at install re_init. EB-2004-12-22-0: Bugfix: The qstat -ext -xml command is broken with N1GE6 Update 2 patch Changed: qstat Issue: 1401 Bugtraq: 6211243 Review: JG MD-2004-12-22-0: Bugfix: uninstall of exec host doesn't work Issue: 1392 Bugtraq: 6211725 Changed: inst_sge Review: HP DT-2004-12-22-2: Bugfix: o Made -no-java the default Changed: aimk JG-2004-12-22-0: Testsuite: Accept new hosts as known hosts when ssh is used as login mechanism. Fixed a timeout problem when closing connections. Fixed error mail (mail body was truncated). Fixed test system_tests/qmaster/size: In runlevel 0 no error condition will be raised when qmaster or schedd show growth. Review: CR DT-2004-12-22-1: Enhancem.: o Updated distinst for DT-2004-12-22-0 Changed: distinst DT-2004-12-22-0: Enhancem.: o Created Ant build script for DRMAA Java language binding Changed: aimk, DRMAA, global Makfile CR-2004-12-21-0: Bugfix: -Werror CFLAG was missing for linux architectures - fixed aimk SG-2004-12-17-0: Enhancem.: - fixed limits for sol10-amd64 Changed: sheperd Review: AA JG-2004-12-17-0: QA: Improved review template and help text. Review: AS JG-2004-12-16-0: Testsuite: Bugfix in performance/throughput check. MD-2004-12-16-0: Bugfix: - permission problem at rc-script gen. Changed: inst_common.sh SG-2004-12-15-0: Enhancem.: - fixed load values and arch for Sol-AMD64 Changed: execd Review: AA SG-2004-12-14-0: Enhancem.: - missed two files in SG-2004-12-13-0 - ported infotext - fixed a output problem in the install Changed: infotext, install, master Review: CR DT-2004-12-14-5: Enhancem.: o DRMAA-J_05 branch is now fully merged Changed: docs, DRMAA DT-2004-12-14-3: Enhancem.: o Merged DRMAA-J_05 docs into main trunk Changed: docs DT-2004-12-13-2: Enhancem.: o Merged DRMAA-J_05 branch into main trunk Changed: DRMAA Issue: 1369, 1370, 1371, 1372, 1373 Review: Pending EB-2004-12-12-0: Enhancem.: Added loadsensor for Interix Review: HP SG-2004-12-13-0: Enhancem.: part 1 of SOLARIS 10 AMD 64. The core system compiles with classic spooling. Changed: everything Review: AA JG-2004-12-09-1: Testsuite: Improved testsuite ssh login. JG-2004-12-09-0: Bugfix: Fixed broken build on lx22-alpha. JG-2004-12-08-1: Testsuite: Suppress output of "washing machine", if testsuite output is not a tty. JG-2004-12-08-0: Testsuite: Fixed infinite loop while waiting for a script file being created. Do not try to shutdown commd processes in 6.x clusters. JG-2004-12-07-2: Testsuite: The java virtual machine to use can now be configured in the host configuration. CR-2004-12-07-0: Bugfix: - error in commlib read/write timeout handling - read/write timeout errors now are reported to the application - removed duplicate function Issue: 1389 Bugtraq: 6205648 Changed: commlib Review: AA DT-2004-12-07-1: Bugfix: o Client commands now fail if GID is unresolvable o qmaster now rejects sharetrees with dangling child references o qconf now rejects charetress with dangling child references Review: EB Changed: qconf, sge_uidgid, sharetree, qmaster, classic spooling Issue: 1256, 1359 Buqtraq: 6205060, 6199256 JG-2004-12-07-1: Testsuite: Fixed installation problems with CSP systems on slow NFS server. JG-2004-12-07-0: Bugfix: Fixed memory leaks in qconf. Changed version string in maintrunk. Fixed clean rule for qmaster Makefile. Review: CR HP-2004-12-06-0: Bugfix: replaced statvfs with wl_statvfs for Interix, fixed DECODE of xdr_double(). Review: EB CR-2004-12-03-2: Cleanup: commlib: - new function: cl_com_get_actual_statistic_data() - test_virtual_qmaster now gets correct nr. of buffered read/write messages - optimizations in TCP framework - qmaster running out of file descriptors (Part I) Issue: 1400 Bugtraq: 6211309 Changed: commlib Review: AA CR-2004-12-03-1: Cleanup: fixed mistake in writing CR-2004-12-03-0: Bugfix: rollback of CR-2004-12-02-3 and CR-2004-12-02-2 adding the dst_len check caused problems, because some invoking function are not setting the correct dst_len. Created an issue to rework replace_params() function. CR-2004-12-02-4: Testsuite: added new format specifier: SFN2 CR-2004-12-02-3: Bugfix: fix of CR-2004-12-02-2 (end of string character would not been written when dst_len is reached) Changed: utilib Review: RH CR-2004-12-02-2: Bugfix: replace_params() in utilib did not check dst_len parameter. Changed: utilib Review: RH CR-2004-12-02-1: Cleanup: - re-adjust qping man page for 2nd qmaster message thread - a reserved port service will only allow reserved port clients - hostname resolve timeouts can now be set via environment variables - general commlib error callback function will now log error strings with a size till 200 chars (was 100 chars before) Changed: qping man page, basis_types.h, commlib, messages Review: MD CR-2004-12-02-0: Bugfix: Fixed broken build for alinux22 HP-2004-12-01-0: Enhancem.: Changes for Windows Port Review: EB CR-2004-12-01-1: Enhancem.: improved performance of lGetElemStrLike() Review: HP SG-2004-12-01-0: Enhancem.: limited the amount of data send to change the busy state in the event client. Review: EB Changed: scheduler, master CR-2004-12-01-0: Enhancm.: qping -dump now can unpack gdi requests and print out the message data Changed: commlib qping Review: AA SD-2004-11-30-0: Bugfix: : Updated security.html with proper information about the kerberos and GSSAPI security code. SG-2004-11-30-0: Enhancem.: ADDED A SECOND GDI THREAD CR-2004-11-26-1: Testsuite: - fixed bug in system_tests/clients/qmake test AS-2004-11-26-0: CVS Created V60u2_TAG --------------------------------------------- CR-2004-11-26-0: Testsuite: - fixed bug in path_alias test which caused errors on irix65 - additional test output for migration test Review: JG DT-2004-11-25-0: Bugfix: o Backed out DRMAA-J changes to create_settings.sh Changed: create_settings.sh MD-2004-11-24-0: Bugfix: arch script does not know about /lib64 Review: JG Changed: arch, arch.dist Issue: 1357 Bugtraq: 6200013 JG-2004-11-24-0: Bugfix: sge_qmaster hit stack limit on aix43, aix51 and hp11 with classic spooling. Review: CR DT-2004-11-24-0: Bugfix: o Fixed segfault in japi_synchronize() Review: AH Changed: JAPI MD-2004-11-23-0: Enhancm.: 60/60u1 to 60u2 BDB update script Changed: inst_sge, inst_update, inst_common, set_fileperm.sh Review: CR AA-2004-11-23-0: Bugfix: qmon segv when deleting sharetree Root node more than once - disallow editing of Root node - disallow changing default node label Issue: 1360 Bugtraq: 6199261 Changed: qmon Review: RH CR-2004-11-23-2: Bugfix: fixed bug in sge_ca script (GE CSP feature): Generating certificate for users failed Changed: sge_ca script Review: AA JG-2004-11-23-0: Testsuite: Verified tests system_tests/geee/sharetree/check.exp and system_tests/utilbin/loadcheck/check.exp. CR-2004-11-23-1: Testsuite: test for issue 1096 does not work with csp system CR-2004-11-23-0: Testsuite: wait 5 seconds after restarting qmaster. It may be a problem for following commands when qmaster is not immediately reachable. JG-2004-11-22-2: Testsuite: fixed check system_tests/shepherd/submit_del Review: CR JG-2004-11-22-1: Testsuite: fixed creation of local configurations (set_config) fixed check system_tests/config/submit_hosts Review: CR CR-2004-11-22-0: Testsuite: fixed remote login testsuite problem Review: JG JG-2004-11-22-0: Bugfix: fixed a memory leak in mirror interface. Review: SG Testsuite: Fixed test system_tests/config/path_alias. Review: CR SG-2004-11-22-0: Bugfix: fixed the scheduler messages for qstat -j. Changed: scheduler Review: EB CR-2004-11-19-1: Testsuite - new test: commlib deadlock test - fixed bug in issue test 1270 MD-2004-11-19-0: Bugfix: Bugfix fix. backup/restore problem. compress/gzip is now supported. backup/restore checks, if host is mastermachine. Changed: inst_common.sh Review: pending AH-2004-11-19-0: Bugfix: Fix DRMAA_DURATION_{H|S}LIMIT misspelled as "durartion" Review: JG Issue: 1345 Bugtraq: 6197253 Changed: drmaa.h, drmaa_attributes(3) JG-2004-11-19-0: Bugfix: Fixed a deadlock in commlib error handling. Review: CR/SG CR-2004-11-19-0: Testsuite: qconf test created files in /tmp/ which can cause problems for multiple testsuites running on the same host. JG-2004-11-18-1: Testsuite: Fixes for tests functional/job_environment and system_tests/clients/qmod, some cleanup AH-2004-11-18-0: Doc: Fix doc/README-Autoinstall.txt missing in distribution. Review: TBD JG-2004-11-18-0: Testsuite: Fixes for tests bugs/issuezilla/1291, system_tests/clients/qalter, system_tests/clients/qsub. Some cleanup in config functions. MD-2004-11-17-0: Bugfix: - backup fails if root is mapped to nobody Review: pending Issue: 1344 Bugtraq: 6196578 Changed: inst_common.sh SG-2004-11-17-0: Bugfix: - the deletion of the queue calendar did not reset the cal for the scheduler. Review: EB Changed: qmaster CR-2004-11-17-0: Testsuite: Fixed some problems for commlib performance tests when grid engine port is < 1024 DT-2004-11-17-0: Bugfix: o Check drmaa_wait() and drmaa_synchronize() for negative timeouts. o drmaa_synchronize() now waits for all tasks ids of all jobs with DRMAA_JOB_IDS_SESSION_ALL parameter. o drmaa_set_attribute() checks validity of DRMAA_TRANFER_FILES value. o drmaa_control() now recognizes held states. Review: AH Changed: japi, drmaa Issue: 1339, 1340, 1341, 1343 JG-2004-11-17-0: Testsuite: Fix and optimization of vi controlling and changing global config. Review: CR AA-2004-11-16-0: Bugfix: - label in Policy Dialog - label width in cluster queue dialog - hostname filtering of jobs Bugtraq: 6195249, 5085010 IZ: 1338 Changed: qmon Review: CR DT-2004-11-16-0: Bugfix: o Increased profiling thread count to 256 o Added validity check for drmaa_transfer_files o Added abort to profiling if max number of threads is exceeded Changed: profiling, DRMAA Review: AH, MD CR-2004-11-16-3: Bugfix: fixed duplicate message id: 33818 MD-2004-11-16-0: Bugfix: backup/restore does not work under Linux and others. Changed: inst_commen.sh Issue: 1328 Bugtraq: 6193866 Review: pending SG-2004-11-16-0: Bugfix: a cluster queue can be deleted, even though it is referenced Changed: qmaster Review: EB Issue: 1316 Bugtraq: 6189289 CR-2004-11-16-2: Enhancm.: - updated qping man page for -dump option - fixed usage output of qping Changed: qping, qping man page CR-2004-11-16-1: Cleanup: removed unused message macros JG-2004-11-16-0: Bugfix: Fixed spooling of subordinate queues. Fixed a memory leak. Review: SG Issue: 1336 Bugtraq: 6194729 CR-2004-11-16-0: Bugfix: replaced u32 format string with U32CFormat in scheduler message text Review: RH Changed: schedd DT-2004-11-15-0: Bugfix: o Fixed typo in id_list_build_from_str_list() o Fixed logic error in qmod o Fixed typo in qmod man page Review: AH Changed qmod, sge_id.c MD-2004-11-15-1: Bugfix: Confusing text Review: HP Changed: inst_qmaster.sh Issue: 791 Bugtraq: 5063317 MD-2004-11-15-0: Bugfix: Confusing text Review: HP Changed: inst_qmaster.sh Issue: 790 Bugtraq: 5063315 HP-2004-11-15-0: Bugfix: Jobs fail on NFS drives mounted without root write privileges - fchown() doesn't work as expected on Irix and Tru64 Review: AH Changed: daemons/common/err_trace.c Issue: 1323 Bugtraq: 6193361 SG-2004-11-15-1: Bugfix: fixed the delivery of start orders (did not work) Review: JG Changed: scheduler SG-2004-11-15-0: Bugfix: removed a ListWrite from the scheduler Review: EB Changed: scheduler JG-2004-11-15-1: Bugfix: Fixed memory leak in qmaster (event_master). Load .sl in hp11-64 instead of .so. Review: SG Updated version string (6.0u2) and GDI version. EB-2004-11-15-0: Bugfix: incorrect select() call in combination with loadsensors Changed: sge_execd Review: JG AH-2004-11-15-0: Bugfix: Use hp11 binaries on HP11 64 bit Use lx24-x86 binaries for Linux 2.6 Review: AS Issue: 1357 Bugtraq: 6200013 AH-2004-11-12-0: Doc: Various fixes with man pages. Review: TBD Bugfix: MD-2004-10-21-0 broke DRMAA Added per thread/per process init to DRMAA setup code Review: DT JG-2004-11-15-0: Bugfix: Fixed broken build. DT-2004-11-12-0: Bugfix: o qmaster now uses very little memory for calculating subordinate queue suspension. o qmaster now ignores NULL job ids. o "" is now rejected as an invalid job id. o qconf now sets status to 1 for failed host group and cluster queue operations. o The lib path will now only be updated if the execd param, SET_LIB_PATH, is set. o The inheritance by jobs of environment of the shell which started the execd can now be disabled with the INHERIT_ENV execd param. o qconf now correctly prints subordinate queue lists. o Subordinate queues are no properly suspended at qmaster startup. o Fixed broken build on Alpha 5 Review: AH, SG, AS, AA, EB Changed: sgeobj, qmaster, execd, shepherd, qconf, qhost Issue: 1289, 1291, 1295, 1300, 1324, 1335 Bugtraq: 6194625, 6176181, 6174915, 6185211, 6193348, 6194713 CR-2004-11-12-0: Bugfix: - commlib called connect() even when connect is already in progress Changed: commlib Review: HP SG-2004-11-12-1: Bugfix: added a qmaster param to set a sharetree spool time Review: DT Changed: qmaster SG-2004-11-12-0: Bugfix: the scheduler made reservations for queues in an unknown state or disabled state. (SG-2004-10-27-0) Review: AH Changed: scheduler JG-2004-11-12-0: Testsuite: performance/throughput: Added scenario with resource reservation. Fixed test system_tests/clients/qmake. Fixed reset_schedd_config CR-2004-11-11-0: Bugfix: sgemaster -migrate on qmaster host starts second qmaster Issue: 1329, 1325 Bugtraq: 6194002 Changed: sgemaster startup script Review: AS JG-2004-11-11-0: Bugfix: Fixed memory access and uninitialized erros in qmaster. Review: DT Testsuite: Minor fixes. CR-2004-11-10-1: Bugfix: sgemaster -shadowd did not start shadowd if no shadowd pid file exist. Review: AA Changed: sgemaster_template CR-2004-11-10-0: Testsuite: - fixed remote loggin problem on slow hosts - increased timeout for csp key installation at execd installation AA-2004-11-10-0: Bugfix: - negative ticket values - default user display in sharetree Issue: 1318 Changed: qmon Review: SG AH-2004-11-10-0: Enhancm.: Improved shepherd diagnosis Review: HP JG-2004-11-10-0: Bugfix: Fixed memory leaks in qmaster. Review: EB Testsuite: Minor fixes. AA-2004-11-09-0: Bugfix: openssl-0.9.7e broke sge_ssl_template.cnf configuration file parsing changed fix in sge_ca and sge_ssl_template.cnf Changed: util/sgeCA/{sge_ca,sge_ssl_template.cnf} Review: CR JG-2004-11-09-0: Testsuite: Fixed test system_tests/clients/qsub. AA-2004-11-08-0: Bugfix: - openssl-0.9.7e upgrade Changed: all Review: CR CR-2004-11-08-0: Bugfix: - fixed qevent -trigger - when qmaster went down once the scheduler always called qmaster alive method in each schedule run - removed sleep time for "qmaster goes down event" - qping -debug now shows also CM/CRM messages - qping -debug connection will get flushed at service shutdown Changed: sge_schedd, qevent, commlib, evclib, mirlib Review: AA JG-2004-11-08-0: Testsuite: Improved error diagnosis of some tests. SG-2004-11-08-0: Bugfix: fixed SG-2004-11-03-0 Changed: master Review: EB CR-2004-11-05-1: Bugfix: fixed broken build CR-2004-11-05-0: Enhancm.: qping -dump (communication traffic) option Changed: qping Review: AA JG-2004-11-05-0: Bugfix: Scheduler didn't respect usage of tightly integrated tasks and logged error messages. Changed: qmaster, scheduler Bugtraq: 6191366 Review: SG SG-2004-11-05-1: Bugfix: added an error message for the case, that a job cannot be started due to the calendar. Changed: scheduler Review: DT SG-2004-11-05-0: Bugfix: the scheduler had critical errors, when host groups were added or modified. Changed: mirror interface, scheduler Review: JG SG-2004-11-04-2: Bugfix: qconf -ks gives bad error message if scheduler isn't running Issue: 1319 Bugtraq: 6201039 Changed: qmaster Review: DT JG-2004-11-04-0: Bugfix: Jobs were not rescheduled to the same host they ran on before. Issue: 1317 Bugtraq: 6201040 Changed: execd Review: SG SG-2004-11-04-1: Bugfix: the scheduler logs 2 error message, when job was deleted by a user during a scheduling run and that you was dispatched from the scheduler. Issue: 1030 Bugtraq: 6201042 Changed: qmaster Review: JG SG-2004-11-04-0: Bugfix: fixed the event master error messages Changed: qmaster Review: DT CR-2004-11-04-0: Enhancm.: commlib: added rresvport() (reserved port) support Changed: commlib Review: AA SG-2004-11-03-0: Bugfix: too many array tasks are deleted Review: EB Changed: qmaster Issue: 1279 Bugtraq: 6190164 AH-2004-11-03-0: Bugfix: qsub doesn't forward the AFS token, removed set_sec_cred call Reschedule unknown blocked host for all jobs of the rescheduled job. Review: AA, SG Changed: scheduler, qsub Issue: 1301 CR-2004-11-03-0: Enhancm.: - commlib debug client support can be enabled via compile time definition "ENABLE_DEBUG_CLIENT" for execd and qmaster daemons. The binary "test_debug_client" can be used as debug client. Changed: commlib, gdilib Review: AA AD-2004-11-03-0: Bugfix: '-m a' qsub option did leave a zombie process Review: EB Changed: qmaster Issue: 1277 Bugtraq: 5104789 CR-2004-11-02-1: Bugfix: - fixed build CR-2004-11-02-0: Enhancm.: - added new commlib test binary: test_debug_client - added new parameter to statistic calculate function to force calculation for qping requests. Otherwise update statistics only once per minute - added debug client support into commlib - fixed bug when shutdown commlib with connected stream clients - fixed possible deadlock in commlib string list when the malloc() of an entry failed. - outcommented unused function: should_flush_event_client() Changed: commlib, gdilib, qmaster Review: AA SG-2004-11-02-0: Bugfix: - fixed a memory leak in the scheduler configuration - fixed a memory leak in the scheduler. It ocoured, when consumables were used as load thresholds. Review: AA Changed: scheduler Issue: 1315 Bugtraq: 6189286 CR-2004-10-29-0: Bugfix: - fixed gdi bug for commlib error reporting - new commlib function: cl_is_commlib_error() Changed: gdi, commlib Review: SG SG-2004-10-29-2: Bugfix: - fixed profiling im scheduler - fixed cpu consumption in event master. The event master never slept. Changed: master, scheduler Review: CR SG-2004-10-29-1: Bugfix: second try on the profiling. Review: JG Changed: scheduler SG-2004-10-29-0: Bugfix: fixed the flush implementation in the event master Review: JG Changed: qmaster CR-2004-10-28-1: Testsuite: commlib throughput test now uses binary_starter.sh to startup more virtual_event/gdi_client binaries parallel JG-2004-10-28-0: Bugfix: qmake did not export shell env vars. Bugtraq: 5090162 Issue: 1313 Review: SG Testsuite: New check system_tests/clients/qmake. Check performance/throughput: Better diagnosis of event delivery interval. Check functional/tight_integration fixed for 5.3. CR-2004-10-28-0: Enhancem.: - code cleanup of commlib - continue work on SSL framework module - removed unused logging output Changed: commlib Review: AA SG-2004-10-28-0: Bugfix: - fixed profiling for the scheduler - fixed the inclusiv/exclusiv profiling of different layers Changed: scheduler, profiling Review: pending AA-2004-10-27-1: Bugfix: qmon returns an error dialog, when editing a calendar Issue: 1267 Bugtraq: 6185169 Changed: qmon Review: CR AA-2004-10-27-0: Bugfix: qmon job customize weird @{} field names additional fields canot be added Issue: 1307 Bugtraq: 6185136 Changed: qmon Review: CR SG-2004-10-27-0: Bugfix: qmod can modify queues starting with digits now. Issue: 1308 Bugtraq: 6184460 Changed: qmod Review: EB SG-2004-10-27-0: Bugfix: - 3/3 part of of the look ahead scheduler extention. Issue: 978 Bugtraq: 6184466 Changed: scheduler Review: EB JG-2004-10-26-0: Testsuite: Additional charts for throughput test: Event delivery interval and number of events delivered. Improved writing of testsuite data, e.g. dump files from throughput test. Fixed a bug in test performance/throughput. CR-2004-10-22-1: Testsuite: moved commlib_throughput test to checktree/performance/commlib CR-2004-10-22-0: Testsuite: - some fixes for new commlib_throughput test SG-2004-10-22-0: Bugfix: - fixed SG-2004-10-21-1 Changed: scheduler Review: JG AH-2004-10-22-0: Doc: Checkpoint README refers to wrong script queue_conf(5) incomplete for owner_list description Issue: 1262, 601 Bugtraq: 6201044, Changed: README.cpr and README.cray JG-2004-10-21-0: Testsuite: Fixed a number of tests, fixed typos and incorrect test descriptions. MD-2004-10-21-0: Enhancem.: - Profiling Enhancement, profiling switchable from the configuration - added profiling to all threads - some cleanup - man pages Changed: qmaster, schedd, execd, profiling, man Review: HP SG-2004-10-21-1: Enhancem.: - 2/3 part of of the look ahead scheduler extention. - code cleanup - minor perf. improvements Changed: scheduler Review: EB SG-2004-10-21-0: Bugfix: qsub und qalter accepted negativ ticket values Issue: 1304 Bugtraq: 6174301 Changed: qsub, qalter Review: EB CR-2004-10-21-1: Testsuite: - new procedure "startup_core_system" - new menu selective "(28) restart cluster" CR-2004-10-21-0: Cleanup: - code cleanup of commlib Review: AA Changed: commlib JG-2004-10-20-0: Bugfix: fixed qsh and test_drmaa (sge_prof_setup) Changed: qsh test_drmaa Review: SG SG-2004-10-20-0: Bugfix: - sge_share_mon crashed on startup Changed: sge_share_mon Review: MD CR-2004-10-19-0: Testsuite: - new testsuite test for commlib: checktree/performance/throughput/commlib/check.exp Changed: commlib test binaries testsuite AH-2004-10-19-0: Doc: Enhanced qstat(1) man page about qstat -j option Changed: man pages Bugtraq: 6181171 Review: AD JG-2004-10-19-1: Bugfix: fixed broken build on alinux JG-2004-10-19-0: Testsuite: Enhanced ssh support: Which ssh client to use is now configured in the host configuration. EB-2004-10-19-0: Bugfix: fix of EB-2004-09-15-0: qstat will show extended information for -pri -ext -urg Issue: not necessary Bugtraq: not necessary Review: SG SG-2004-10-18-0: Enhancem.: - reworked the locking in qmaster and changed two not thread save code sections in the order processing - moved the persitence of the job sequ. number into the timed event thread - add a copy date flag to the GDI-Multi request Changed: qmaster, scheduler, clients Review: EB CR-2004-10-18-0: Cleanup: - code cleanup of commlib - removed unused function parameters in list functions - preparatory work for commlib SSL framework Review: AA Changed: commlib utilib AH-2004-10-18-0: Bugfix: meaningless job error state diagnosis text in qstat -j Review: JG Changed: execd Issue: 1299 Bugtraq: 6180529 JG-2004-10-15-0: Testsuite: Fixed csp installation, installation of SGE(EE) 5.3, package generation for SGEEE 5.3. JG-2004-10-14-0: Testsuite: Testsuite now can create tar packages after compiling by setting the package_type parameter to "create_tar". CR-2004-10-13-0: Enhancem.: - implemented application status callback function for execd. - qping man page update - CR-2004-10-07-1 (final Part III) - additional code cleanup - faster execd reconnect to qmaster when qmaster was restarted Review: AA Changed: qmaster execd utilib commlib gdilib Issue: 1294 Bugtraq: 6176115 JG-2004-10-14-0: Bugfix: Fixed broken build. EB-2004-10-13-0: Enhancem.: qstat requests only necessary jobs (qstat -s -u) Changed: qstat, qhost, libuti, sge_qstat(5), qstat(1) Review: SG HP-2004-10-13-0: Bugfix: Checked in wrong sources yesterday Changed: sge_uidgid, windows library HP-2004-10-12-0: Enhancem.: Changed build output path and binary path for Windows version, added 'Administrator' detection. Changed: sge_uidgid, windows library JG-2004-10-12-0: Bugfix: Fixed broken build on AIX. SG-2004-10-12-0: Bugfix: The qmaster crashed every once in a while when processing the peding orders. Introduced in SG-2004-09-27-0. Changed: scheduler Review: pending SD-2004-10-11-0: Enhancem.: added user to unable to run interactive job log message Changed: qmaster Issue: 1296 MD-2004-10-11-2: Bugfix: fixed segfaults of clients Changed: qdel, qalter, qevent, qacct, qhost SG-2004-10-11-0: Enhancem.: Part 1/3 of the look ahead scheduler extention. - code cleanup - every job has a duration now. Changed: scheduler Review: EB JG-2004-10-11-0: Enhancem.: performance tuning - improved Berkeley DB throughput - cleanup of qmaster - execd protocol Review: CR MD-2004-10-11-1: Bugfix: fixed qmod (sge_prof_setup) Changed: sge_profiling CR-2004-10-11-0: Bugfix: removed unnecessary logging output Review: HP Changed: commlib MD-2004-10-11-0: Bugfix: fixed broken build Changed: sge_profiling MD-2004-10-08-1: Bugfix: restoring a backup does not restore the job_scripts dir Review: CR Changed: inst_sge Issue: 1239 Bugtraq: 6176177 MD-2004-10-08-0: Enhancem.: Profiling Code Enhancement (made MT-save) Review: CR Changed: qmaster, schedd, sge_profiling Issue: Bugtraq: CR-2004-10-08-0: Enhancem.: implemented application status callback function for qmaster. CR-2004-10-07-1 (Part II) Review: MD Changed: qmaster utilib Issue: 1294 Bugtraq: 6176115 CR-2004-10-07-1: Enhancem.: application status callback function now supports status string (Part I) Review: AA Changed: commlib Issue: 1294 Bugtraq: 6176115 DT-2004-10-07-0: Bugfix: o Removed slots from complex values list of queue instances in qconf o Enabled qsub/qrsh -v VAR again o Fix job exit error message o Added missing (not spec'ed) BUFFER to drmaa.h o Added job category spec Review: CR Changed: qconf, flatfile spooling, qmaster, qsub/qrsh Issue: 1198, 1255, 1287 Bugtraq: 6174326, 6174331, 5086108 CR-2004-10-07-0: Cleanup: qping -help now shows short Grid Engine Version name Changed: qping CR-2004-10-05-2: Cleanup: added some comments Changed: commlib header file cl_xml_parsing.h CR-2004-10-05-1: Testsuite: added missing install expect handling for qmaster installation EB-2004-10-05-0: Enhancem.: added new module test which especially tests the modifications done with EB-2004-09-15-0 Changed: none CR-2004-10-05-0: Bugfix: segmentation fault when vmemsize limit is reached o removed unnecessary logging information in commlib o prepare_enroll() will now terminate application on cl_com_setup_commlib() errors Review: AA Changed: commlib, gdi Issue: 1286 Bugtraq: 6174821 DT-2004-10-04-0: Bugfix: o Fixed testsuite test for 1269 o Execd now also looks in $SGE_ROOT/bin for qidle and qloadsensor o qconf is now able to handle unbreakable lines longer than MAX_STRING_SIZE Changed: testsuite, execd, flatfile spooling Review: SG Issue: 1176, 1251 Bugtraq: 5108635, 5108639 CR-2004-10-01-0: Bugfix: Thread enabled commlib coredumps on exit on a 32bit Solaris x86 box Issue: 1283 Bugtraq: 5075968 Review: MD Changed: commlib JG-2004-09-29-0: Bugfix: Automatic users were deleted even if they had jobs (running or pending). Review: SG Changed: qmaster SG-2004-09-29-0: Bugfix: the master spooled the user and projects in every scheduling run Review: JG Changed: qmaster JG-2004-09-28-0: Enhancem.: Cleanup in qrsh_starter. Cleanup and speedup in load report processing. Add hashtables on cluster queue and queue instance attributes. Review: SG SG-2004-09-27-0: Tuning: improved ticket order generation Changed: qmaster, scheduler Review: JG JG-2004-09-23-0: Bugfix: Fixed broken build. CR-2004-09-23-0: Bugfix: more precise error messages for communication errors Review: JG Issue: 1275 Bugtraq: 5097732 Changed: commlib, gdilib CR-2004-09-22-0: Cleanup: - removed unused messages - qmaster will now respond faster to qconf -km - added some comments to the source code Review: JG EB-2004-09-21-0: Bugfix: qmaster might fail if jobs are deleted which have multiple hold states applied Issue: 1280 Bugtraq: Review: SG CR-2004-09-21-0: Testsuite: added a cleanup function for test bugs/issuezilla/1270 JG-2004-09-20-0: Testuite: Improved test performance/throughput, added a number of charts to comparison of two clusters. Review: CR SG-2004-09-20-0: Enhance.: First part of extending the scheduler to use the calendars. The qi-data structure has been extended to include the next 2 state changes, and how long the current and the next last. The calendar implementation had to be changed to allow that. Changed: qmaster, test_calendar Issue: 1235 Bugtraq: 5104270 Review: EB JG-2004-09-17-0: Enhancem.: Performance improvement of the bitfield implementation. Review: SG Issue: 1260 DT-2004-09-15-1: Bugfix: o Fixed qmaster crash o Added leftover files from DT-2004-09-15-0 - Testsuite test for 1269 not yet working Changed: sge_host.c Review: SG Issue: 1269 Bugtraq: 5102442 EB-2004-09-15-0: Enhancem.: qstat requests only that attributes of JB_Type and JAT_Type elements which are really needed to execute qstat successfully Changed: all Review: SG SG-2004-09-15-1: Bugfix: - fixed a locking issue in the loging. Changed: qmaster Review: CR Issue: 1268 Bugtraq: -NA- JG-2004-09-15-0: Bugfix: Fixed a bug in event master. Changed: qmaster Review: SG DT-2004-09-15-0: Bugfix: o Fixed drmaa_synchronize() problem Changed: JAPI Review: SG Issue: 1270 Buqtraq: 5102340 SG-2004-09-15-0: Bugfix: fixed memory leak in the scheduler Changed: scheduler Review: DT Issue: 1274 Bugtraq: 5102320 DT-2004-09-14-0: Bugfix: o Fixed DRMAA-J README o Fixed JavaDoc error in aimk o Updated DRMAA-J JavaDocs Changed: docs, aimk JG-2004-09-14-0: Testsuite: Improved test performance/throughput: - Monitoring of qmaster communication (qping). - Additional senarios: - run without qstat - jobs with resource requests - parallel jobs with resource requests - array jobs - functional policy Review: CR DT-2004-09-13-1: Bugfix: o Fixed DRMAA-J on Irix 6.5 DT-2004-09-13-1: Bugfix: o Added libjdrmaa to LD_LIBRARYN32_PATH in settings.sh|csh for Irix 6.5 Changed: create_settings.sh DT-2004-09-13-0: Bugfix: o Fixed "delete all" bug in JAPI o Fixed locking bugs in JAPI o Tested DRMAA-J on all platforms - No J2SE 1.4 on Tru64 5.0 and AIX 4.3 o Fixed attribute getter error in DRMAA-J o Added drmaa.jar to CLASSPATH in settings.sh|csh Changed: JAPI, DRMAA, aimk, distinst, create_settings.sh Review: EB SG-2004-09-13-0: Bugfix: fixed memory leak in the scheduler Changed: scheduler Review: DT MD-2004-09-13-0: Testsuite: - fixed install scripts message handling - added new macro JG-2004-09-09-0: Testsuite: Improved test performance/throughput: - monitor size and cpu usage of qmaster and schedd - do qstat calls during the test run and monitor the qstat runtime and errors - verified, fixed and documented the calculated metrics. Some testsuite cleanup. Review: CR SG-2004-09-08-0: Cleanup: - fixed a bug added by SG-2004-09-08-0 - got the test_sge_qmaster_timed_event working - cleaned qmaster_main Changed: qmaster Review: AA SG-2004-09-08-0: Cleanup: - improved the locking in the event master - filters now for subscribed events Changed: qmaster Review: JG SG-2004-09-08-0: Cleanup: improved the loging Changed: all Review: AA JG-2004-09-07-0: Testsuite: Improved test functional/tight_integration (different sizes of massive parallel jobs are tested) SG-2004-09-06-0: Cleanup: Cleanup in the order processing, event master, and event client Changed: qmaster, scheduler Review: EB AH-2004-09-06-0: Bugfix: Effect of -p priority and weight_priority not described in sge_priority(5) Changed: man page Issue: 1265 Bugtraq: 5097481 Review: JG AA-2004-09-06-1: Bugfix: parsing of repeated job arguments fixed Bugtraq: 5097479, 6185208 Issue: 1263, 1302 Review: RH Changed: qmon AA-2004-09-06-0: Bugfix: make culltrans compile under 6.0 Changed: culltrans Review: RH EB-2004-09-06-0: Buffix: qacct -l is not working Review: SG Issue: 1261 Bugtraq: 5095907 Changed: qacct JG-2004-09-06-0: Testsuite: Enhanced / fixed test performance/throughput: - reduced performance impact of testsuite monitoring on the Grid Engine performance - additional runlevels for additional scenarios and higher numbers of jobs - fixed computation of cluster utilization Did some fixes in testsuite procedures: - installation of 5.3 systems - filedescriptor leak in chart creation Review: CR JG-2004-09-01-0: Enhancem.: Cleanup and performance improvement for bitfields. Enhanced the bitfield module test. Review: SG Issue: 1260 HP-2004-08-31-0: Bugfix: Fixed some minor bugs: wrong format specifier, Typo in man page, ... Changed: man-page, commlib-testprogram SG-2004-08-30-0: Bugfix: new algorithm for o-ticket Changed: man-page, scheduler Review: MD Issue: 1224 Bugtraq: 5094016 JG-2004-08-30-0: Enhancem.: Added module test for libs/uti/sge_bitfield. SG-2004-08-27-0: Bugfix: fix SG-2004-08-19-0 crashed the qmaster, if qsub -w v was submitted Changed: scheduler HP-2004-08-26-1: Bugfix: Fixed scripts/zerodepend for No-Windows version HP-2004-08-26-0: Enhancem.: Added 'stubs' for Interix (Windows) Version SG-2004-08-26-0: Bugfix: hard resource requests ignored in parallel jobs Review: AH Issue: 1253 Bugtraq: 5092487 Changed: scheduler CR-2004-08-26-0: Cleanup: - code cleanup - added debug code for issue #1197 - code is not active (#if 0 /#endif) - added comments CR-2004-08-25-1: Cleanup: - removed all #ifndef ENABLE_NGC code and removed makro definition in aimk CR-2004-08-25-0: Bugfix: - added man pages for qping, gethostname, gethostbyname, gethostbyaddr, getservbyname - the binaries now support -help option and show also the GE version number. Review: SG, DT Issue: 1132 Bugtraq: 5071878 JG-2004-08-25-0: Bugfix: Fixed build on AIX CR-2004-08-23-0: Testsuite: Fixed mail report error in tight_integration_test MD-2004-08-20-0: Bugfix: merged bugfix from FCS_Branch Review: AS Changed: inst_sge, inst_common.sh, rc_templates CR-2004-08-20-0: Testsuite: enhancement: tight integration test checks execd messages files for errors Issue: 1199 CR-2004-08-19-1: Enhancem.: added debug information about GDI messages: don't show debug info when no message was received. EB-2004-08-19-0: Bugfix: Submit to a queue domain is never scheduled Review: SG Issue: 1234 Bugtraq: 5089255 Changed: sge_schedd CR-2004-08-19-0: Enhancem.: added debug information about GDI messages Changed: gdi lib Review: RH SG-2004-08-19-0: Bugfix: scheduling weirdness with wild-card PE's Changed: scheduler Review: DT, AH Issue: 1216 Bugtraq: 5089222 CR-2004-08-17-3: Cleanup: - removed unused message macros CR-2004-08-17-2: Enhancem.: - added new function to transmute the gdi tag id into a string: sge_dump_message_tag() CR-2004-08-17-1: Enhancem.: - added comment about flushing job reports when jobs are started in execd. CR-2004-08-17-0: Testsuite: - added wait time in compile_source - raised wait time for qmaster shutdown AS-2004-08-17-0: CVS Created V60u1_TAG --------------------------------------------- AA-2004-08-12-0: Bugfix: - qmon_help.ad help corrections - sge_centry_qconf.c missing parentheses break build - qmon_calendar.c NONE strings were missing for added calendars - qmon_ckpt.c Reschedule flag is part of Signal When flags - qmon_cluster.c delegated_file_staging flag added - qmon_manop.c set corresponding tab when invoked from Policy dialog dialog resize - qmon_util.c hostgroups with @f* were not displayed Issues: reported during tests Review: CR Changed: qmon MD-2004-08-12-0: Bugfix: changed wrong directory of qping binary autoinst shadowd Changed: inst_sge CR-2004-08-12-0: Testsuite: raised timeout when waiting for qmaster to get down RH-2004-08-12-0: Bugfix: Refix Issue 1161 Issue: 1161 Bugtraq: 5080856 Review: CR Changed: qconf CR-2004-08-11-4: Testsuite: - fixed bug which caused error in qmod reschedule test - fixed tcl bug in issue test 193 MD-2004-08-11-1: Bugfix: - sge backup cannot override older backups Issue: 1222 Bugtraq: 5085508 Review: HP Changed: inst_sge MD-2004-08-11-0: Bugfix: - sge_inst restore does not work Issue: 1219 Bugtraq: 5085507 Review: HP Changed: inst_sge CR-2004-08-11-3: Testsuite: - fixed bug in procedure wait_for_unknown_load: endless loop if do_error_check is set to 0 DT-2004-08-11-1: Bugfix.: o Rolled back dangerous fix from DT-2004-07-16 Review: SG CR-2004-08-11-2: Testsuite: - usage test did not work with only 2 execd hosts - wait_for_unknown_load call from shutdown_core_sytem did not return immediately when no binaries are installed or when qconf returns an error DT-2004-08-11-0: Doc.: o Cleaned up man3, man8, and submit.1 man pages Issue: 1170 Review: DT CR-2004-08-11-1: Testsuite: - fixed bug which caused error in issue test 314 CR-2004-08-11-0: Testsuite: - fixed bug in issue test 193 MD-2004-08-10-0: Bugfix: Refix Issue 1113 Issue: 1214 Bugtraq: 5084927 Review: HP Changed: inst_sge CR-2004-08-10-2: Testsuite: - wait for all queues to report load before starting any test - wait for all queues to report unknown load after qconf -ke all in shutdown_core_system CR-2004-08-10-1: Testsuite: startup_shadowd did not use sgemaster -shadowd command line argument CR-2004-08-10-0: Testsuite: shutdown_core_system will now wait for qmaster shutdown when qconf -km is used CR-2004-08-09-0: Bugfix: fixed fix for CR-2004-08-05-2 execd may exit when hostname resolving is changed or not available. Review: AA SG-2004-08-09-0: Bugfix: the reporting file was not written correctly under linux on opteron. Issue: 1207 Bugtraq: 5084317 Review: AA CR-2004-08-06-0: Bugfix: Wrong entry SGE_QMASTER_PORT in qconf man page Issue: 1112 Changed: qconf man page Review: AA CR-2004-08-05-2: Bugfix: Need more verbose diagnosis msg if execd port is already bound Issue: 1101 Bugtraq: 5083115 Changed: commlib, execd, gdi, japi Review: SG AA-2004-08-05-0: Bugfix: tight integration doesn't work with CSP Issue: 1080 Bugtraq: 5063314 Changed: all Review: CR EB-2004-08-05-0: Bugfix: hostgroup changes do not always take effect. Issues: 1185 Bugtraq: 5083102 Changed: qmaster Review: SG CR-2004-08-05-1: Bugfix: killed slave task on tight integration does not cleanup job correctly (part 1) Issue: 1197 Bugtraq: 5082660 Changed: execd Review: AH CR-2004-08-05-0: Testsuite: unsetting SGE_DEBUG_LEVEL in testsuite test scripts pe_task.sh and pe_job.sh SG-2004-08-04-1: Bugfix: fixed an aspect of bug 1195: the error message truncated to 100 characters. Review: RH Issue: -NA- Bugtraq: -NA- Changed: qsub SG-2004-08-04-0: Bugfix: qstat -ext -urg did not display the submit/start time Review: RH Issue: 1192 Bugtraq: 5082490 Changed: qstat RH-2004-08-04-0: Bugfix: Missing global defintion of variable check_root_access_needs in some bugfix tests added Changed: testsuite (tests for issues 1096 1104 1141 1146 and 1161) Review: CR Issue: -NA- Bugtraq: -NA- SG-2004-08-03-2: Bugfix: fixed qsat -g c output and qmon to show cqload instead of load. The man page for qstat was changed as well. Review: AA Changed: qstat, qmon, qstat.1 Issue: 1181, 1190 Bugtraq: -NA- EB-2004-08-02-0: Bugfix: Deleting a queue instance slots value actually adds it Issue: 1175 Bugtraq: 5081822 Changed: qmaster Review: SG SG-2004-08-03-1: Bugfix: fixed qconf -ahgrp to work with a given name. It will generate a templet, that has to be modifed by the user. Issue: 1186 Bugtraq: 5081839 Changed: qconf, qmaster Review: DT SG-2004-08-03-0: Bugfix: fixed a bug in the xml-otuput Issue: 1191 Bugtraq: 5081821 Changed: qstat Review: EB DT-2004-08-01-0: Bugfix: o Fixed JavaDoc errors o Fixed qstat man page error o Updated aimk to build new JavaDocs o Updated distinst to install drmaa.jar in lib o Added a link to the JavaDocs from the DRMAA Java language binding howto Changed: doc, aimk, distinst, www DT-2004-07-30-0: Bugfix: o Fixed seg fault in qconf -mattr o Fixed CULL cross-linking in qconf -mattr queue complex_values o Fixed hostname aliasing issue in qconf -mattr o qhost now prints NCPU as an int o Changed qsub -r to accept 'yes' and 'no' o Cleaned up JAPI Issue: 1055, 1081, 1092, 1172 Bugtraq: 5080836, 5080839, 5080833, 5080840 Changed: qconf, qhost, qsub, JAPI, doc Review: SG SG-2004-07-30-1: Bugfix: qconf -de did not check if the host ist still in use. Issue: 1183, 1168 Bugtraq: 5080779, 5080784 Changed: qstat, qmaster Review: DT HP-2004-07-30-0: Doc: Description of exit_status values of shepherd Issue: 97 Changed: man8/sge_shepherd.8 SG-2004-07-30-0: Enhancem.: added a feature to filter the jobs in the scheduler based on their category. Changed: schedler, schedd.conf(5) Review: AD AD-2004-07-30-0: Cleanup: The implementation of 'sge_add_auto_user()' made the use of recursive mutexes neccessary. Changed: qmaster Review: SG SG-2004-07-29-0: Bugfix: Resending queue signals broken Changed: exed, qmaster Issue: 1178 Bugtraq: 5079572 Review: EB CR-2004-07-28-1 Bugfix: - Startup of qmaster changes act_qmaster to `hostname`, - Two network cards in exec host: creates wrong directory / shutdown fails Bugtraq: 5071522, 5079514 Issue: 1110, 1113 Changed: install scripts, shadowd, gdi lib Review: RH JG-2004-07-29-0: Bugfix: qrsh starter didn't output reason (strerror), if it couldn't write its pid file. Testsuite: Make test functional/tight_integration use all hosts of the testcluster. Use category VERIFIED as default. Changed: qrsh_starter, testsuite Review: CR AA-2004-07-28-1: Bugfix: qmon's queue filtering doesn't work Bugtraq: 5075936 Changed: qmon Review: RH AA-2004-07-28-0: Bugfix: Wallclock time limit in qmon Bugtraq: 5078783 Issue: 1162 Changed: qmon Review: CR SG-2004-07-28-0: Bugfix: - fixed the event master theading problem - fixed the job names and accouting string evaluation. - removed an abort in qmaster, when sched conf was not valid. Bugtraq: 5075849 Issue: 1099 Changed: qmaster, submit.1 Review: JG DT-2004-07-27-0: Enhancem.: o Finished enhancing DRMAA Java language binding docs o Checked in JavaDocs for DRMAA Java lang. binding Bugfix: o Corrected error in DRMAA README o runBulkJobs now returns all ids o Minor edits to DRMAA man pages Changed: DRMAA, docs Issue: 1165 JG-2004-07-27-0: Bugfix: Fixes in reporting file: - pe_task_id for non pe jobs were inconsistend (in some cases "none", in others "NONE"). - Intermediate accounting records contained invalid data for submission_time and start_time. Changed: sge_qmaster Bugtraq: 5078872 Bugfix: Fixed a memory leak when reading from Berkeley DB. Review: RH DT-2004-07-26-1: Bugfix: o Fixed minor regex issue with findcheckin.pl Changed: scripts CR-2004-07-26-0: Testsuite: fixed shutdown of shadowd problem (testsuite does not shutdown shadowd) Issue: 1166 Review: JG RH-2004-07-26-2: Bugfix: definition of wc_cqueue added to man page. Issue: 1103 Bugtraq: Not defined Changed: sge_types(1) Review: AH DT-2004-07-26-0: Enhancem.: o Added script to locate files by CVS log header Changed: scripts RH-2004-07-26-1: Bugfix: man page indicated that qsub accepts "-r yes|no" Issue: 1103 Bugtraq: ???? RH-2004-07-26-0: Bugfix: Wrong parameter for -Mc option argument qconf(1) man page fixed. Enhancements in man page for sched_conf(5). In man pages submit(1) the '$' and '.' as forbidden characters for job names removed. Issue: 1114, 969, ??????? Bugtraq: ????, ???, 5076358 Changed: qconf(1), sched_conf(5), submit(1) Review: SG SG-2004-07-26-0: Bugfix: - first part of the event_master changes. The total update of a event client is now thread safe. - excluded the flowwing character from a job name: "\" Changed: qmaster, submit.1 Review: JG SG-2004-07-25-0: Bugfix: - fixed lWriteListTo to use the spedified file handle - some minor cleanup - changed the job name specification that all characters are allowed except: "\n", "\t", "\r", "/", ":", "@" Issue: -NA- Bugtraq: Changed: qmaster, submit.1 CR-2004-07-24-1: Bugfix: schedd and qmaster get out of sync - no scheduling for long time Changed: communication lib Issue: 1141 Bugtraq: 5077589 Review: JG CR-2004-07-23-0: Testsuite: modified /functional/tight_integration check to be more error tolerant Review: AA MD-2004-07-22-2: Bugfix: reprioritize_interval descr in sched_conf(5) needs improvement Changed: man5/sched_conf.5 Issue: 1040 Bugtraq: 5077165 Review: JG MD-2004-07-22-1: Bugfix: NO_REPRIORITATION should be removed from man pages Changed: man5/queue_conf.5, man5/sge_conf.5, sge_priority.5 Issue: 1097 Bugtraq: 507167 Review: JG MD-2004-07-22-0: Bugfix: uninstall procedure doesn't remove rc-script of execd! Changed: inst-sge, autoinstall template Issue: 1130 Bugtraq: 5071868 Review: JG DT-2004-07-21-1: Bugfix: o Added "-w e" as default DRMAA option o Fixed message number conflict o Fixed segfault in qconf -mc o Added test suite test for qalter -h and -t (Issue 1109) o Added test suite test for Issue 1146 o Added test suite test for Issue 1161 o Cleaned up JAPI o Fixed broken links on DRMAA-J howto page Changed: DRMAA, qmaster, testsuite, qconf, www Issue: 1109, 1146, 1151, 1161, 1163 Bugtraq: Review: SG CR-2004-07-21-0: Testsuite: modified /functional/tight_integration check to be more error tolerant Review: JG JG-2004-07-21-0: Bugfix: Fixed a bug in aimk -shared-libs. Fixed default busy handling in event client. Doc.: Described use of bdb_checkpoint.sh in Berkeley DB README file. Review: CR DT-2004-07-21-0: Enhancem.: o Completed DRMAA Java language binding howto o Updated DRMAA Java language binding README o Started enhancing DRMAA Java language binding JavaDocs Changed: DRMAA, www AD-2004-07-20-0: Bugfix: calendars broken Issue: 1111 Bugtraq: 5071502 Changed: qmaster Review: SG SG-2004-07-20-0: Bugfix: jobs on hold due to -a time cause qmaster/schedd get out of sync Issue: 1154 Bugtraq: 5074788 Changed: scheduler Review: DT MD-2004-07-20-0: Bugfix: variable syntax : equal sign support Issue: 1118 Bugtraq: 5075398 Review: JG Changed: qsub, qrsh, qsh MD-2004-07-19-0: Bugfix: sched_conf(5) reprioritize_interval should default to 0 Issue: 1139 Bugtraq: 5075451 Review: SG Changed: schedd default config, schedd DT-2004-07-19-1: Bugfix: o Changes from DT-2004-07-16-0 review Changed: DRMAA, qstat, qconf, qsub AD-2004-07-19-1: Bugfix: 'qconf -km' did cause surplus warning. Changed: qmaster SG-2004-07-19-0: Bugfix: - The usage in the sharetree got lost. - fixed a mirror interface time out, which disabled sheduling intervals greater than 1 min. - fixed a mt problem on qmaster shutdown Changed: qmaster Issue: 1094 Bugtraq: 5075346 Review: DT JG-2004-07-19-0: Testsuite: Fixed testuite setup. AD-2004-07-19-0: Bugfix: Did fix broken qmaster locking Review: SG Changed: qmaster DT-2004-07-19-0: Enhancem.: o Added code examples for DRMAA Java language binding howto. (Done over the weekend) Changed: DRMAA DT-2004-07-16-0: Bugfix: o Removed chdir() from DRMAA job processing o Made -cwd an (optionally) forbiddon DRMAA switch o Fixed minor DRMAA Java binding bugs o Fixed double-counting of used, unavailable queue slots o Changed qstat -g c and qmon to show np_load_avg instead of load_avg o Fixed error reporting for invalid characters in queue names in qconf o Allowed -w e and -w n as valid DRMAA switches o Fixed abort in qdel, et al with invalid -t swtich o Removed "NONE" from qconf -aconf o Fixed typo: WARINING_VERIFY => WARNING_VERIFY Changed: DRMAA, qstat, qmon, qconf, qdel, qmod, qalter, qsub, qmaster Issue: 1150, 1125, 1127, 1146, 1151, 1109, 1137 Bugtraq: 5072005, -, -, 5073218, -, 5071525, - Review: SG AH-2004-07-13-0: Bugfix: Deleted pending job appears in qstat Problem was that finished jobs generally were displayed by qstat. Removed broken code that was supposed to mask out running/finished jobs at qmaster side. Added code in qmaster to improve diagnostics in cases such as new found bug #1154. Issue: 1117 Bugtraq: 5072481 Review: JG MD-2004-07-13-0: Bugfix: inst_sge -sm doesn't create a local_conf Issue: 1135 Bugtraq: 5071999 Review: CR Changed: testsuite, inst_sge AA-2004-07-13-0: Bugfix: removed stat_log_time from qmon Issue: 1083 Review: CR Changed: qmon AH-2004-07-13-0: Bugfix: Revert change AH-2004-07-09-0. Bug #1117 is not fixed. Issue: 1117 Bugtraq: 5072481 Review: CR RH-2004-07-13-0: Bugfix: qmod -e '@' causes segmentation fault in qmaster in the method qref_resolve_hostname (libs/sgeobj/sge_qref.c) a sprintf "@%s" instead of "%s@%s" is executed if the queuename part is empty. Issue: 1131 Bugtraq: 5071918 Review: JG JG-2004-07-13-1: Bugfix: fixed build for some gcc versions. JG-2004-07-13-0: Testsuite: fixed test system_tests/geee/sharetree CR-2004-07-13-0: Testsuite: - modifications for 60 scheduler configuration in throughput test Changed: throughput test Review: JG CR-2004-07-12-0: Bugfix: - cl_com_tcp_connection_complete_request() got a new parameter to handle only "read", only "write" or "read and write" connection establisment. ( read and write thread did both and produced a one second timeout for multithreaded clients. Now the read thread will only read and the write thread will only write. A non multithreaded commlib will do both, read and write ) - new function cl_commlib_search_endpoint(): The function will return an endpoint list with connected clients. - moved some functions to new module: cl_xml_parsing.c Changed: commlib Review: HP JG-2004-07-12-0: Bugfix: qmaster spooled tightly integrated parallel tasks whenever online usage was reported by an execd. Added ADOC header for job/ja_task/pe_task spooling. Changed: sge_qmaster Issue: 1129 Bugtraq: 5072772 Review: AH SG-2004-07-12-0: Bugfix: - qstat -j -xml printed unparsable xml-output - the qstat xml schema missed the hard request queue field. Changed: qstat, detailed_job_info.xsd Issue: 1144 Bugtraq: -NA- Review: AA AH-2004-07-09-0: Bugfix: Deleted pending job appears in qstat Changed: qstat, qhost Issue: 1117 Bugtraq: 5072481 Review: CR SG-2004-07-09-0: Bugfix: scheduler ignores queue seqno for queue sorting Changed: scheduler Issue: 1133 Bugtraq: 5071914 Review: RH CR-2004-07-07-0: Bugfix: qping doesn't support host_aliases file Changed: qping binary Issue: 1124 Bugtraq: 5071539 Review: AA MD-2004-07-08-0: Bugfix: Qmaster requires a local conf in order to start Issue: 1104 Bugtraq: 5071987 Review: CR Changed: qmaster_configuration JG-2004-07-07-0: Testsuite: - Source documentation for compile procedure. - Disable caching of testsuite generated html files. JG-2004-07-06-0: Bugfix: Fixed the build for HPUX. CR-2004-07-06-0: Bugfix: qmaster clients may not reconnect after qmaster outage Changed: communication library (commlib) Issue: 1126 Bugtraq: 5076491 Review: JG JG-2004-07-05-1: QA: Update of review form and help text. Review: CR JG-2004-07-05-0: Bugfix: projects not available after sge_qmaster restart Changed: sge_qmaster Issue: 1122 Bugtraq: 5071498 Review: CR DT-2004-07-05-0: Bugfix: o Fixed broken build on AIX o Fixed typos in JGrid Changed: shepherd, JGrid JG-2004-07-01-0: Testsuite: Added check for qmaster spooling containing a testcase for issue 1122. EB-2004-06-23-0: Cleanup: Added/Changed ADOC commets EB-2004-06-22-0: Cleanup: Added ADOC commets CR-2004-06-18-1: Testsuite: - new check for CR-2004-06-18-0 (issues 1096 and 1098) - new configuration parameter: reserved_port - fixed testsuite problem on shutdown of multi threaded qmaster process(es!) on linux systems. CR-2004-06-18-0: Bugfix: o sge_qmaster will not start unless run only as "root" o qmaster cannot bind port below 1024 on Linux Changed: qmaster Issue: 1096, 1098 Bugtraq: 5063987 DT-2004-06-17-0: Bugfix: o Fixed bad link in DRMAA howto (again) o Added copyright placeholders to Java source files o Renamed DRMAA Java binding README o Fixed various JGrid bugs o Fixed factory bug in DRMAA Java binding o Fixed multi-threaded errno issue on HP-UX 11 o Added JGrid to experimental html file - HP11 and HP1164 will need a aimk -clean Changed: doc, www, classes, DRMAA, JGrid, aimk MD-2004-06-17-0: Bugfix: remove stat_log_time from configuration Changed: inst_qmaster.sh Review: AS Issue: 1082 Bugtraq: 506330 MD-2004-06-16-0: Bugfix: Install script fails when sgeadmin is selected as install user Changed: inst_qmaster.sh Review: JG Issue: 1090 SG-2004-06-16-0: Bugfix: This bugfix is related to SG-2004-06-09-2. The prefious one did not fix all aspects of the growing scheduler and master. The implementation of the job categories was buggy. Changed: scheduler Review: JG Issue: 1087 Bugtraq: 5063311 DT-2004-06-15-1: Bugfix: o Fixed DRMAA Java binding Makefile o Fixed installation of libjdrmaa in distinst Cleanup: o Removed tabs from aimk help output o Removed unused code from sge_path_alias.c Enhancem.: o Ported JGrid to DRMAA Java binding Changed: aimk, distinst, sge_path_alias, DRMAA, JGrid DT-2004-06-15-0: Bugfix: o Fixed minor bugs in DRMAA Java binding o Fixed bad link in DRMAA howto Enhancem.: o Added Java binding README and started howto Changed: DRMAA, howtos JG-2004-06-14-0: QA: Updated review template and doc. Cleanup: Removed unused I18N messages. Testsuite: Make checks performance/throughput and system_tests/qmaster/size check for IZ 1087 DT-2004-06-08-0: Bugfix: o Backed out AS-2004-06-03-0 for DRMAA README. Changed: docs SG-2004-06-09-3: Enhancem.: updated doc/devel/review.txt Review: JG SG-2004-06-09-2: Bugfix: The scheduler did not corectly compress the scheduler info messages, which lead to a hugh memory consumption in bigger clusters with even a low amount of jobs in the system. Changed: scheduler Review: JG Issue: 1087 Bugtraq: 5063311 SG-2004-06-09-1: Bugfix: - fixed minor issues in the scheduler profiling Changed: scheduler Review: EB SG-2004-06-09-0: Bugfix: - fixed the build for Linux with spool-classic Changed: aimk Review: JG JG-2004-06-09-0: QA: Added development howto's: - use of branches - procedure for fixing bugs JG-2004-06-08-0: QA: Added review template and help text. AS-2004-06-08-0: CVS Created V60_TAG_{BRANCH,TAG} DT-2004-06-04-1: Bugfix: o Fixed DRMAA Java binding build on Linux o Added per thread init to japi_exit o Fixed exit bug in TestDRMAA o Removed old JNI files from JAPI directory o Removed trailing garbage from German file staging howto Changed: WWW, DRMAA DT-2004-06-04-0: Enhancem.: o Added German translation of file staging howto o Enhanced TestDRMAA Changed: WWW, DRMAA DT-2004-06-03-5: Bugfix: o Fixed API error in DRMAA Java binding Enhancem.: o Expanded TestDRMAA program Changed: DRMAA MD-2004-06-03-0: Bugfix: typos, some messages, cleanup Changed: inst_sge, setfileperm.sh Review: AS AS-2004-06-03-1: Cleanup: changed message when "qmod -X" mod is not done because queue already has the specified state. AS-2004-06-03-0: Bugfix: formatting, typos of various text files DT-2004-06-03-4: Bugfix: o Fixed check-in problem with DT-2004-06-03-3 Changed: DRMAA DT-2004-06-03-3: Bugfix: o Fixed various bugs in DRMAA Java binding Changed: DRMAA JG-2004-06-03-0: Testsuite: Improved tests functional/access_lists. Cleanup. MD-2004-06-02-1: Bugfix: - added comment to autoinst template Changed: inst_template.conf DT-2004-06-02-2: Bugfix: o Fixed DRMAASession.wait() in DRMAA Java binding o Switched Java binding to use libjdrmaa.so Changed: classes Issue: 1079 MD-2004-06-02-0: Bugfix: - changed message in scheduler settings. Changed: inst_sge DT-2004-06-02-1: Bugfix: o qconf no longer presents a default project when editing projects Cleanup: o Created separate Makefile for DRMAA Java binding Changed: DRMAA, flatfile spooling, aimk, distinst Issue: 1077 JG-2004-06-02-0: Bugfix: QU_full_name of queue instances was not set after a qmaster restart. Review: AD DT-2004-06-02-0: Cleanup: o Updated DRMAA Java binding JavaDocs Changed: DRMAA AD-2004-06-01-0: Bugfix: Did add workaround for non POSIX compliant behaviour of LinuxThreads regarding propagation of UID/EUID to new threads. Changed: qmaster MD-2004-06-01-1: Bugfix: - permission problem, rc-scripts creation Changed: inst_sge, inst_berkeley.sh DT-2004-06-01-3: Bugfix: o Fixing DT-2004-06-01-2 Changed: qsub Issue: Related to 1075 JG-2004-06-01-1: Bugfix: Fixed memory leaks in japi and drmaa libs. Review: DT DT-2004-06-01-2: Bugfix: o Fixing DT-2004-06-01-1 - NOTE: checking in *so* I can compile and test Changed: qsub Issue: Related to 1075 DT-2004-06-01-1: Bugfix: o Better handling of errors in qsub - NOTE: checking in *so* I can compile and test Changed: qsub, JAPI Issue: Related to 1075 DT-2004-06-01-0: Bugfix: o Fixed state of JAPI event client thread on aborted startup. o Fixed typo in howtos Changed: JAPI, howtos Issue: 1075 MD-2004-06-01-0: Bugfix: Db reinitialized during restore Changed: inst_sge JG-2004-06-01-0: Testsuite: Don't set file permissions when installing zip packages. Fixed qmaster size test. SG-2004-05-28-0: Bugfix: fixed qmaster on HP-11. The problem was a "char save[32k]" in the file: daemons/qmaster/sge_give_jobs.c Changed: qmaster Review: MD MD-2004-05-28-3: Enhancem: README-Upgrade.txt DT-2004-05-28-5: Bugfix: o Resolved conflict between new file staging howto and previous Changed: howtos MD-2004-05-28-2: Bugfix: - message Changed: inst_sge MD-2004-05-28-1: Enhancem: - auto backup Changed: inst_sge DT-2004-05-28-4: Bugfix: o Added file staging to howto main page Changed: howtos Issue: 952 DT-2004-05-28-3: Bugfix: o Changes to DRMAA howto o Added file staging howto Changed: howtos Issue: 952 MD-2004-05-28-0: Bugfix: - some file creation/deletion - hints - logging Changed: inst_sge JG-2004-05-28-1: Bugfix: Fixed install_execd and berkeley db rpc server installation. Review: MD DT-2004-05-28-2: Cleanup: o Documented issue in sge_flatfile.c Changed: Flatfile spooling DT-2004-05-28-1: Bugfix: o Clarified drmaa_wait() man page Changed: docs DT-2004-05-28-0: Bugfix: o Fixed DRMAA C binding revision number Changed: DRMAA JG-2004-05-28-0: Bugfix: Fixed memory leaks in fileparsing and in japi. Enhancem.: Reduced stack memory usage of static dstrings. Review: SG, DT MD-2004-05-27-1: Bugfix: - backup/restore directory check - berkeley info text - rc template qmaster shutdown Review: AS, JG JG-2004-05-27-2: Bugfix: Fixed memory leaks in execd. Review: SG AA-2004-05-27-0: Bugfix: - qmon_about mailto text - CSP reconnect problem Changed: all DT-2004-05-27-0: Bugfix: o Added DRMAA example programs to distinst o Updated DRMAA README Changed: distinst, docs Issue: 952 Review: AS JG-2004-05-27-1: Bugfix: Fixed memory leaks in scheduler. Review: SG AH-2004-05-27-0: Bugfix: Deadline scheduling broken Issue: 1063 Review: DT JG-2004-05-27-0: Bugfix: Fixed memory leaks in scheduler. Fixed a core dump in qmaster shutdown. Review: SG MD-2004-05-27-0: Bugfix: - uninstall order changed - added message - fixed usage Changed inst_sge DT-2004-05-26-0: Bugfix: o Moved DRMAA README into docs directory and added it to distinst Changed: docs Issue: 952 MD-2004-05-26-2: Bugfix: - improved logging Changed: inst_sge JG-2004-05.26-2: Bugfix: Fixed memory leaks in qmaster and in commlib. Review: SG SG-2004-05-26-1: Bugfix: The scheduler did sometimes dispatch more jobs into a queue, than that queue had resources left. This only happend in advanced reservation mode. Review: AH, DT Changed: sge_schedd MD-2004-05-26-2: Bugfix: -csp option does not recognize secure binaries Review: AS Changed: inst_sge Issue: 1065 MD-2004-05-26-1: Bugfix: -some small path problems Review: RH Changed inst_sge JG-2004-05-26-1: Enhancem.: Add script and utility for checkpointing and cleanup of transaction log in case of Berkeley DB spooling with RPC server. Review: MD MD-2004-05-26-0: Bugfix: - refixed file perms setting of /tmp/configuration - MoveLog procedure Review: RH Changed: inst_sge SG-2004-05-26-0: Bugfix: fixed a scheduler crash. The scheduler crashed when the host settings were not straight forward, and not all load values were delivered. Review: AA Changed: scheduler JG-2004-05-26-0: Bugfix: Fixed several memory leaks in qmaster. Review: SG JG-2004-05-25-1: Testsuite: Fixed testsuite installation in csp mode. BD-2004-05-25-0: Bugfix: - Use -ldb$(BDB_LIB_SUFFIX) instead of bare -ldb in utilbin/Makefile. AA-2004-05-25-0: Bugfix: - man page sched_conf - qmon_help.ad - Q helpbox visible lines Issue: 448 Changed: qmon, doc MD-2004-05-25-1: Buxfix: - delete spool dir, message was not shown - setting file perms to /tmp/configuration - asking for SGE_ROO/SGE_CELL during uninst, if not set Changed: inst_sge DT-2004-05-25-0: Bugfix: o Fixed build for Solaris make o Changed DRMAA job usage to unscaled Changed: DRMAA, execd Makefile SG-2004-05-25-2: Bugfix: removed a compiler warrning in event master Changed: qmaster Review: JG AD-2004-05-25-1: Bugfix: Did fix operator precedence issue. Changed: qmaster SG-2004-05-25-1: Bugfix: fixed a memory leak in qmaster Changed: qmaster Review: JG SG-2004-05-25-0: Bugfix: fixed a memory leak in qmaster Changed: qmaster Review: JG, DT JG-2004-05-25-0: Testsuite: Fixed installation (parsing messages from inst_sge). Include $SGE_CELL in testsuite mail headers. MD-2004-05-25-0: Bugfix: added many log messages Changed: inst_sge Review: pending AD-2004-05-25-0: Bugfix: Configuration version increment has been lost. Changed: qmaster Review: pending MD-2004-05-24-0: Bugfix: - Berkeley DB RPC server is running as user root - inst_sge -db doesn't install rc script (typo) - typos Changed: inst_sge Issue: 1060, 1059 Review: AA DT-2004-05-24-0: Bugfix: o Cleaned up flatfile spooling error codes o Cleaned up qconf error messages o Reworked JAPI initialization to delay GDI calls until necessary o Expanded DRMAA Java binding test program o Renamed DRMAA README o Added -help_arch to distinst o Updated DRMAA man pages Changed: qconf, flatfile spooling, JAPI, DRMAA Issue: 1005, 1025 Review: AH, SG SG-2004-05-24-1: Bugfix: The monitor switch in the scheduler would not turn off, after removing the monitor entry from the params in the scheduler config. Changed: scheduler Review: JG SG-2004-05-24-0: Bugfix: Fixed qstat -j ID output Changed: qstat Issue: 898 Review: AD JG-2004-05-24-2: Bugfix: qdel of non existing job/array task/task range as admin user produced an incorrect error message. Review: SG Cleanup: Removed outdated Insure instrumentation. Testsuite: Adapted testsuite to changed qdel error messages. JG-2004-05-24-1: Cleanup: Removed unused I18N messages. JG-2004-05-24-0: Testsuite: Add $SGE_CELL to testsuite mails. Fixed remote execution of testsuite commands. JG-2004-05-23-0: Testsuite: Additional changes required for multi cell support. Fixed installation using RPC server. Removed unused testsuite code, macros and scripts. AD-2004-05-21-0: Bugfix: If possible use 'sysconf()' to determine group buffer size. If not use increased default value. Issue: 960 Changed: uti Review: pending AA-2004-05-21-0: Bugfix: halflife and compensation factor are not i refrshed on qmon compiler complaints IZ: 1058 Changed: qmon Review: pending JG-2004-05-21-1: Testsuite: Testsuite now supports multiple cells. JG-2004-05-21-0: Bugfix: - Event master didn't send list events (function sge_add_list_event). - Fixed error in reaper_execd (could reap wrong ja_task). - Fixed build on lx??-*. Changed: qmaster, execd Review: AA DT-2004-05-21-0: Bugfix: o Backed out DT-2004-05-14-0 and DT-2004-05-14-1 o Made reaper search the entire job list for tasks Changed: execd Review: pending Issue: 1031, 1034 BD-2004-05-19-5: Enhamcem: - If 3rdparty/{qmake,qtcsh}/COMPILE_ARCH does not exist, create it using configure. This should be vastly more maintainable then the current checked in copies, but will not interfere with them. BD-2004-05-19-4: Enhamcem: - Add three new variables to aimk.site and use them in aimk and Makefiles to allow Berkeley DB's headers and libraries to live in subdirectories of include/ and lib/ as well as allowing the library to have a suffix. This is needed to allow FreeBSD and NetBSD to build against the DB version from ports/pkgsrc. BD-2004-05-19-3: Bugfix: - Revert incomplete rename of RC_FILE value "freebsd" to "fbsd" BD-2004-05-19-2: Enhamcem: - FreeBSD-amd64 support BD-2004-05-19-1: Enhamcem: - allow compilation on FreeBSD BD-2004-05-19-0: Cleanup: Clean up configure related files in qmake and qtcsh so configure generates useful output. Changed: qmake, qtcsh AD-2004-05-19-0: Bugfix: - Did eradicate global 'Master_Config_List'. Qmaster configuration module now does use a module local list, which is properly synchronized. - Did streamline Qmaster configuration module interface. Issue: 1041 Changed: qmaster, execd, evm, sgeobj, spool Review: pending JG-2004-05-19-1: Bugfix: Fixed broken qdel -f output (see also IZ 664). Cleanup: Removed job_log from execd and scheduler. Changed: qmaster, schedd, execd Review: SG MD-2004-05-19-0: Bugfix: - Upgrade Welcome Text - inst_template comments - autoinst always used default schedd config Changed: inst_sge AA-2004-05-19-0: Bugfix: - warning when execd is started as non-root user - make csp work with admin user daemons - change of motif lib for pippin Issue: 857 Changed: all bins Review: pending CR-2004-05-19-0: Bugfix: o Removed -lj Option for execd o added/fixed help output to sge_execd, sge_qmaster sge_shepherd and sge_coshepherd (IZ#1026). o fixed bug in when host_aliases are active (all mapped host names had to be resolvable on local host, now mapped names must not be real resolveable host names ) o removed duplicate error message when execd can't create communication handle (bind tcp port) Issue: 1026 Changed: qmaster, execd, shepherd, coshepherd, libcom Review: JG DT-2004-05-19-0: Bugfix: o Fixed errors in DRMAA howto Changed: howto JG-2004-05-19-0: Bugfix: Qmaster logged errors when immediate array job couldn't be scheduled. Issue: 1037 If qmaster was shutdown before finished jobs were deleted via schedd order, these jobs showed up again as running after qmaster restart. Changed: qmaster Review: SG DT-2004-05-18-0: Cleanup: o Cleaned up some code and comments o Added test program for DRMAA Java binding o Cleaned up JNI code for DRMAA Java binding Changed: qsub, DRMAA, various others MD-2004-05-18-4: Enhamcem: - more comments in inst_template.conf JG-2004-05-18-3: Bugfix: Fixed a core dump in qmaster. SG-2004-05-18-4: Bugfix: fixed the qstat -t -xml output. Issue: 1043 Changed: qstat Review: JG SG-2004-05-18-3: Bugfix: changed the qdel man page Changed: qdel.1 MD-2004-05-18-3: Bugfix: - backup/restore works (only Berkeley DB spooling) - help output Changed: inst_common.sh Issue: 1003, 1004 SG-2004-05-18-2: Bugfix: changed the qdel behavior back to the v5.3 one. Review: JG Changed: qmaster Issue: 1042 JG-2004-05-18-2: Bugfix: qmaster Master_* lists weren't deleted at shutdown. Fixed a core dump in qmaster if reporting file couldn't be opened. Free the spooling context at qmaster shutdown. Review: SG MD-2004-05-18-2: Bugfix: - forgot sourcing of a module Changed: inst_sge MD-2004-05-18-1: Bugfix: -Help output -changed recommended switches Changed: inst_update.sh MD-2004-05-18-0: Bugfix: log file in wrong directory, autoinst failed under AIX -docu Changed: inst_sge, docu Issue: 984, 996 Review: AS (pending) Enhancem: SGE 6.0 upgrade procedure Changed: inst_update.sh CR-2004-05-18-0: Bugfix: fixed strstr() call to not zero terminated string Changed: commlib Review: JG JG-2004-05-18-1: Bugfix: fixed memory leaks in qmaster. Changed: sge_qmaster Review: SG SG-2004-05-18-1: Bugfix: fixed memory leak in qmaster. Changed: sge_qmaster Review: JG SG-2004-05-18-0: Bugfix: fixed memory leak in qmaster. Changed: sge_qmaster Review: JG JG-2004-05-17-0: Bugfix: Fixed a memory leak in qmaster. Changed: sge_qmaster Review: SG AA-2004-05-17-0: Bugfix: - AIX51 load, switch on incomplete AIX51 load code with aimk -DTEST_AIX51 - make broken on AIX43 IZ: 754 Changed: sge_execd, load_check Review: CR SG-2004-05-14-0: Bugfix: - fixed the messages generation for jobs, which exceeded the user job limit. - added copyright statements to the xml-schemas Review: RH, AH Changed: scheduler, XML schemas DT-2004-05-14-2: Bugfix: o Fixed core dump in event master Changed: EVM Review: pending CR-2004-05-14-1: Bugfix: Can't interrupt qacct -j with Ctrl-C Issue: 981 Review: AA Changed: qacct AH-2004-05-14-1: Bugfix: qstat -j shows wrong message for parallel jobs which can't be dispatched Partial fix. Issue: 814 Review: SG JG-2004-05-14-1: Testsuite: test functional/migration was broken due to changed qstat queue output. AH-2004-05-14-0: Bugfix: Change of consumable values takes long time until used by scheduler. Issue: 1033 Review: JG JG-2004-05-14-0: Bugfix: Fixed build dependencies for spooledit and japi howtos. DT-2004-05-14-1: Bugfix: o Fixed build Changed: execd CR-2004-05-14-0: Bugfix: qacct: strange 'B' charcater after maxvmem value Issue: 824 Changed: qacct, objlib Review: SG DT-2004-05-14-0: Bugfix: o Fixed the way the last element is removed from a range o Execd now reuses existing job structures for array job tasks o Fixed character printing problem in lWriteElem o Updated DRMAA README o Updated DRMAA Java binding o Checked in missing Java binding class o Miscelaneous cleanup Changed: range, execd, CULL, JAPI, DRMAA Issue: 1031, 1034, 1035 Review: SG, JG JG-2004-05-13-2: Bugfix: - for jobs that did no resource requests, the accounting record contained an empty field (::) which was incorrectly parsed by qacct. - spooldefaults failed reading exechosts that have complex variables defined. - aimk clean didn't delete japi howto object code. JG-2004-05-13-1: Testsuite: testsuite install re_init failed. SG-2004-05-13-0: Bugfix: scheduler died due to the use of a none reduced job descriptor. Changed: scheduler Review: JG JG-2004-05-13-0: Cleanup: Cleanup for nightly build: - fixed duplicate message id's - removed unused messages - create libdrmaa and test programs on AIX43 JG-2004-05-12-3: Testsuite: Adapted testsuite qstat parsing to the extended queue column from SG-2004-05-11-1 (IZ 1016). AA-2004-05-12-0: Bugfix: - qstat -l a=sol-sparc64 job display fix - wrong message during install Issue: 1002 Changed: qstat, install_script Review: SG CR-2004-05-12-0: Bugfix: o Repeated commlib logging when qmaster can't be reached o Fixed problem of checkin by CR-2004-05-11-1: PID before sge_daemonize() is not the same after daemonize. The /tmp/execd_messages.PID file wasn't appended to execd/messages file. Issue: 864 Changed: execd, gdilib Review: SG JG-2004-05-12-2: Enhancem.: Added sharetree to the spooldefaults targets for the update procedure. JG-2004-05-12-1: Bugfix: Reentered dependency fix introduced to aimk by JG-2004-05-07-0 - it got lost again in one of the following checkins. SG-2004-05-12-1: Bugfix: updated the qstat man page Changed: qstat.1 EB-2004-05-12-0: Bugfix: qdel of just submitted job Issues: 126 Changed: sge_execd Review: JG JG-2004-05-12-0: Testsuite: Fixed test system_tests/clients/qmod/reschedule. Worked around parsing unfriendly output of qstat -g t for parallel jobs. SG-2004-05-12-0: Cleanup: cleaned the xml schema and the qstat xml output to be consitent. Changed: qstat, qstat.xsd Review: MD DT-2004-05-11-0: Enhancem.: o Finished DRMAA Howto Changed: howto SG-2004-05-11-1: Bugfix: qstat prints queue names with a max length of 30 per default. SGE_LONG_QNAMES can be used to set a custom length (min 10) or -1. In case of -1 qstat computes the length for the queue names. The man pages are still missing. Issue: 1016 Changed: qstat Review: AA CR-2004-05-11-2: Bugfix: o L10N for sge_get_qmaster_port() and sge_get_execd_port(). o sge_get_qmaster_port() and sge_get_execd_port() will now cache resolved ports. Issue: 1036 Changed: utilib Review: JG SG-2004-05-11-0: Bugfix: - fixed that user are spooled after every scheduling run, when the sharetree is used. The usage in the user are updated every 2 min now. - reseted the ticket amount for jobs, which are in hold, or error state. These jobs are not scheduled and should not keep their old tickets. - If a immediate array job was not scheduled, qsub returned that the job was scheduled. Issue: 980, 604 Review: DT Changed: scheduler JG-2004-05-11-1: Testsuite: Fixed test system_tests/config/file_parsing. AA-2004-05-11-0: Bugfix: - Startup logging contains 'none' - qhost does not print overwritten ARCH - qmon policy dialog, missing show pending job tickets Issue: 861, 813 Review: EB Changed: qhost, qstat, qmon EB-2004-05-11-0: Bugfix: Changed msg id JG-2004-05-11-0: Bugfix: If the CULL definition of an object was extended by additional attributes without changing the spooled fields, the object was unspooled incorrectly by a newly built qmaster. Bugfix: qacct failed if jobs were started with a very long job name. Review: SG CR-2004-05-11-1: Bugfix: o removed ignored parameters from prepare_enroll() o execd now appends it's pid to temporary startup file name ( /tmp/execd_messages.PID) o execd will now exit after a timeout when binding SGE_EXECD_PORT fails o communication lib errors are now L10N enabled Changed: commlib gdilib Issue: 1023, 767 Review: JG CR-2004-05-11-0: Bugfix: sge_gdi_setup() with NULL as alpp can cause sgementation fault. Review: AA EB-2004-05-10-1: Bugfix: requestable=forced is ignored for built-in complex(5) attributes Changed: sge_schedd Issue: 936 Review: AH, SG DT-2004-05-10-0: Enhancem.: o Began work on DRMAA howto, including example programs o Cleaned up qsub Changed: howto, JAPI, qsub AD-2004-05-10-0: Cleanup: - First part of 'Master_Config_List' encapsulation. - 'lWriteElem_' and 'lWriteList_' now use 'stderr' is no 'FILE*' argument is given. Changed: qmaster, cull, sgeobj, exced Review: SG EB-2004-05-10-0: Bugfix: jobs submitted with option -r n can be rescheduled Changed: qalter, qsub Issue: 999 Review: DT JG-2004-05-10-0: Testsuite: Fixed test system_tests/qmaster/size. CR-2004-05-10-0: Bugfix: commlib: o cl_commlib_receive_message() will now trigger communication estabilshment when connection list is empty. The application will not have to react (sleep/trigger) this case when it acts as a service. o fixed test_cl_endpoint_list application o enhancements to test_virtual_xxx applications CR-2004-05-07-0: Bugfix: commlib: o forbid to create a service handle with component id 0 o when a new connected client doesn't write any data (is just listening) after connection establishment a read timeout appeared at service component and the connection was closed because of read timeout error. o added code to ensure that an application will not go into an endless loop when a connection can't be opened (endless application calls to cl_commlib_receive_message() in a while loop without guilty communication endpoint). Changed: commlib DT-2004-05-07-1: Bugfix: o Added test to test_drmaa for invalid attributes o Did cleanup related to 966 o Added Java DRMAA binding to aimk and distinst - use aimk -jdrmaa to include the Java binding in libdrmaa.so and to build drmaa.jar - distinst ignores drmaa.jar if it can't find it Changed: DRMAA, qsub, JAPI, aimk, distinst Issue: 966 DT-2004-05-07-0: Bugfix: o Removed duplicate qconf error messages o Fixed build on MacOS X o Fixed qsub bug introduced in DT-2004-04-06-0 Changed: qconf, sgeobj, qsub Issue: 868 Review: SG JG-2004-05-07-0: Bugfix: Calling dbstat while qmaster was running caused a spooling error in qmaster. Fixed dependencies to spooling libs (qmaster, spoolinint and spooldefaults were linked with every aimk call. Enhancem.: Allow dbstat to write objects - renamed it to spooledit. Change dump format from lWriteElem to lDumpElem. Install spooledit to utilbin. Issue: 762 Changed: qmaster, spooledit (dbstat), libspool* Review: AD CR-2004-05-06-0: Enhancem.: added new test binaries to setup a communication test: test_virtual_event_client, test_virtual_gdi_client and test_virtual_qmaster DT-2004-05-06-0: Bugfix: o Fixed bad error messages for qconf -mc o Fixed return code for qsub when max jobs reached o Improved qconf error messages o Updated DRMAA man pages o Fixed byte ordering bug in JGrid Changed: sgeobj, qconf, centry, qsub, JGrid Issue: 773, 868, 1013 Review: SG AA-2004-05-06-0: Bugfix: qtcsh shows strange error messages on AIX 43/51 Issue: 849 Changed: qtcsh Review: CR JG-2004-05-05-1: Bugfix: Fixed parsing of report_variables in qconf. It now uses a comma separated list of variables. Adjusted man page accordingly. Check if entries listed as report_variables exist as complex variables. Prevent complex variables from being deleted while they are still referenced in report_variables. Allow the setting of default report_variables in the global host object. Changed: qmaster, qconf Issue: 1008 Review: EB SG-2004-05-05-1: Bugfix: event master: - event ack are processed before a client can deregister - a client will not get events, after it got a shutdown event - in case of job_mod and cqueue_mod events are the sub-lists (ja_tasks / qinstance) removed from the event. Review: DT Changed: qmaster, scheduler CR-2004-05-05-2: Testsuite: debug_starter.sh did not set SGE_ND CR-2004-05-05-1: Bugfix: o qmaster not responding (no client will receive a gdi response) o setting max possible component id to 65535 Changed: commlib Issue: 1022 SG-2004-05-05-0: Enhancem.: disabled hashing for cull list which are prepared for sending em to a client. Changed: sge_qmaster Review: JG JG-2004-05-05-0: Testsuite: Verified tests bugs/issuezilla/171, bugs/issuezilla/403, functional/migration, system_tests/clients/qalter. Fixed call of qdel -u "*". Increased timeouts for qmaster shutdown. EB-2004-05-04-1: Bugfix: qrsh jobs are not scheduled to interactive queue Changed: sge_schedd Issue: 855 Review: JG JG-2004-05-04-1: Bugfix: Removed debug output to stderr from qstat/qhost. JG-2004-05-04-0: Enhancem.: spooldefaults handles additional objects required for the update procedure (adminhosts, calendars, ckpts, cqueues, exechosts, projects, submithosts, users) Review: MD EB-2004-05-04-0: Bugfix: qstat -g c slot counting broken Changed: qstat, qmon Issue: 975 Review: SG CR-2004-05-03-1: Enhancem.: o modified code to speed up connection establishment Changed: commlib CR-2004-05-03-0: Enhancem.: o cl_commlib_receive_message() is now only triggered, when a new message for the application is ready for read. Changed: commlib Review: SG SG-2004-05-03-0: Enhancem.: changed the order in the order package that is send from the scheduler to the qmaster. Changed: scheduler Review: AH SG-2004-05-03-0: Bugfix: fixed a bug in the mirror interface, which did accept events, after the client shutdown event. This let to one error message per event. The real fix needs to be done in the event master. Changed: scheduler Review: JG JG-2004-05-03-1: Testsuite: Fixed test system_tests/clients/qmod/reschedule (fails due to IZ 999). JG-2004-05-03-0: Enhancem.: Optimization of hashtables. Review: SG DT-2004-05-02-0: Enhancem.: o Started DRMAA README o Cleaned up DRMAA web page o Checked JGrid into experimental directory o Fixed byte order issues in JGrid skeleton Changed: DRMAA, JGrid Issue: 952 CR-2004-05-01-0: Bugfix: o global host list refresh (re-resolving of used host names) is now done in a separate list copy to be able to unlock the list and let other threads use the global host list in the meantime Changed: commlib AD-2004-04-30-1: Bugfix: Did fix build. Changed: qmaster DT-2004-04-30-0: Bugfix: o drmaa_control(DRMAA_CONTROL_TERMINATE) now behaves like qmon when deleting o Brought test_drmaa in sync with drmaa.c - test_drmaa still fails the DRMAA_V_ENV test in ST_ATTRIBUTE_CHECK. Changed: DRMAA Issue: 979 AD-2004-04-30-0: Cleanup: Did move 'get_local_conf_val()'. Changed: qmaster JG-2004-04-30-1: Testsuite: verified test functional/access_lists fixed test system_tests/clients/qmod/auto_reschedule JG-2004-04-30-0: Bugfix: qconf -sconf had problems with short vs. long hostnames and hostname resolution. qmaster did no hostname resolving for cluster config objects. deleting config or host objects that had a long hostname by specifying the short hostname and vice versa failed. Fixed a qmaster core dump. Changed: sge_qmaster, qconf Issue: 1007, 1009 Review: SG CR-2004-04-29-2: Enhancm.: - removed unused structure element in cl_com_hostent_t structure - added new functions: cl_com_copy_in_addr(); cl_com_copy_hostent(); cl_host_list_copy(); Changed: communication Review: SG SG-2004-04-29-1: Bugfix: changed the locking in the com. connection list Changed: communication Review: CR SG-2004-04-29-0: Bugfix: the event master would only deliver a limited number of events. In case of a scheduler, this results in wrong scheduling runs. Changed: qmaster Review: DT JG-2004-04-29-0: Bugfix: Fixed classic spooling of jobs. Changed: qmaster Review: EB Issue: 1006 CR-2004-04-29-1: Bugfix: fixed core dump of qmaster when setting a host to unheard. SG-2004-04-29-0: Bugfix: qdel -u "*" jid deleted all jobs of all users. Changed: qmaster Review: JG Issue: 1000 CR-2004-04-29-0: Enhancm.: Added more help output to qping application JG-2004-04-28-2: Testsuite: Fixed tests bugs/issuezilla/423 and system_tests/clients/qconf (add_user/mod_user). CR-2004-04-28-0: Bugfix: - fixed problems when file descriptor limit is reached. (mark execd unheart) - raised nr. of file descriptors reserved for qmaster ( e.g. spooling) - fixed bug when a connection is opened from two service endpoints at the same time (e.g.: qmaster and execd at) - threads which are unkown to communication lib (e.g. qmaster message thread) didn't log into communication log list - qping didn't shutdown commlib on error. JG-2004-04-28-1: Cleanup: Removed no longer used directories QUEUE_DIR and COMPLEX_DIR. Review: MD JG-2004-04-28-0: Testsuite: Fixed tests system_tests/clients/qdel and system_tests/clients/qsub, use $SGE_ROOT/util/resources/loadsensors/ibm-loadsensor on AIX machines. DT-2004-04-28-0: Bugfix: o Fixed deadlock in EVM from DT-2004-04-27-0 Changed: o EMV Review: SG DT-2004-04-27-1: Enhancem.: o Further progress on DRMAA Java binding Changed: DRMAA Java binding AD-2004-04-27-0: Enhancem: Did add support for 'gcov' tool on LINUX and DARWIN Changed: aimk Review: AH SG-2004-04-27-0: Bugfix: - fixed locking problem between the message and the event thread - enabled the gdi - event transactions. Every gdi request is one transaction. Changed: qmater Review: DT, CR EB-2004-04-27-2: Bugfix: Wrong qconf -sq CQ@HOST output for xuser_lists Issue: 991 Changed: qconf Review: JG EB-2004-04-27-1: Bugfix: scheduling info should say "queue instance" Issue: 965 Changed: sge_schedd JG-2004-04-27-2: Testsuite: Added support for architectures irix65 and darwin. Made DRMAA test use runlevel "long". DT-2004-04-27-0: Bugfix: o Added missing SGE_ROOT error message in japi_init() o Broke up job deletes into 500 job chunks in japi_control() and japi_exit() o Added check for valid attribute in drmaa_set_attribute() o Fixed handling of $drmaa_wd_ph$ in DRMAA o Fixed handling of rusage parameter in drmaa_wait() o Fixed memory leak in qsub o Further progress on DRMAA Java binding o Fixed locking bug in EVM o Optimized locking in EVM send_events o Fixed handling of gdi_setup error messages in clients o All DRMAA calls now check to be sure the session exists before continuing, either via japi_init_was_called() or the regular JAPI calls. - test_drmaa currently fails due to a commlib issue - test_drmaa needs to be updated to check return codes for drmaa_allocate_job_template() and drmaa_set_attribute() Issue: 661, 774, 872, 909, 986, 988, 990 Changed: JAPI, DRMAA, GDI, all clients, EVM Review: AH, SG JG-2004-04-27-1: Testsuite: Extended test bugs/issuezilla/126. Cleanup in several tests. Fixed setting of local qmaster spool directory in SGE 6.0 installation check. EB-2004-04-27-0: Bugfix: qstat -F short cut does not work Cleanup: register variables Issue: 204, 886 Changed: qstat, qhost Review: RH, SG JG-2004-04-27-0: Bugfix: Disable dynamic loading of spooling libs on IRIX65, AIX43 and AIX51: The classic spooling references global variables (Master_*_List) and reverse resolution of symbols in DSO's seems to be broken on these platforms. Changed: qmaster, spoolinit, spooldefaults Review: EB MD-2004-04-26-0: Bugfix: qping fails if, SGE_QMASTER_PORT was not set. (In case of using sge_qmaster service) Changed: sgemaster_template Review: AH Issue: 989 EB-2004-04-26-0: Enhancem.: ADOC-comments Cleanup JG-2004-04-23-0: Bugfix: Fixed core dump in event master module. Changed: qmaster Review: AD AD-2004-04-22-1: Bugfix: Changed thread startup order back to old one. Register job resend events for jobs with status 'JTRANSFERING' in a separate function. Changed: qmaster Review: AH SG-2004-04-22-2: Bugfix: - fixed memory leak in berkeleydb spooling for jobs - fixed memory leak in qsub Changed: qmaster, qsub Review: AS, DT SG-2004-04-22-1: Bugfix: fixed memory leak cluster queue modification Changed: qmaster Review: DT, EB HP-2004-04-22-0: Enhancem.: Added Docu for delegated file staging JG-2004-04-22-1: Bugfix: qmake output warning "cannot determine architecture" even if the commandline contained an explicit resource request. Now read from environment variable SGE_ARCH instead of ARCH. Review: patch submitted by uddeborg Issue: 983 Bugfix: fixed aimk.site for HP11 Review: AS Bugfix: Fixed automatic installation. Review: MD AA-2004-04-22-0: Bugfix: - loadadjustments display of load values was broken Changed: qmon Review: SG SG-2004-04-22-0: Bugfix: - fixed memory leak in sge_event_master - added a cleanup method for all lists on qmaster shutdown - scheduler would exit, if the qmaster is gone for a while Changed: qmaster, schedd Review: CR, AD MD-2004-04-22-0: Bugfix: LD_LIBRARY_PATH was wrong. -> wrong ARCH string job check during uninstall, if rescheduled or no Review: JG Changed: rc_templates CR-2004-04-22-0: Bugfix: qping now also supports auto close when qmaster runs out of filedescriptors DT-2004-04-22-1: Bugfix: o Minor bugs fixes to event master o Added DRMAA test case to test suite o Fixed qconf bug about whitespace after the queue name with -mq Changed: EVM, GDI, test suite, qconf AD-2004-04-22-0: Bugfix: Changing the log level did only affect the thread which did trigger the change. Changed: uti Issue 953 Review: SG JG-2004-04-22-0: Enhancem.: Allow the mixed use of binaries with dynamic and static spooling. aimk.site contains a rule to switch to static spooling for platforms that don't support dynamic loading of libraries (HP11). Review: AS, MD AH-2004-04-22-0: Bugfix: Minor fix with usermapping(5) Review: RC DT-2004-04-22-0: Enhancem.: o Further update to DRMAA Java binding Changed: DRMAA Java binding (maintrunk) AH-2004-04-21-0: Bugfix: Fix broken qalter -R y|n Issue 982 Review: AD CR-2004-04-21-0: Bugfix: Ensure that max. nr of dynamic event clients never exceeds max. nr of file descriptors used for communication Changed: qmaster Review: SG AS-2004-04-21-0: CVS Created V60_beta2_target_{BRANCH,TAG} ----------------------------------------------------------------- DT-2004-04-21-0: Enhancem.: o Streamlined event master locking o Did various cosmetic fixes o Added script for submiting multiple simultaneous qsubs Changed: EVM, cull, qmaster, GDI, JAPI, mirror, UTI, scripts Review: SG, AD MD-2004-04-21-2: Bugfix: removed confusing text! Changed: rc-template JG-2004-04-21-0: Enhancem.: distinst: Copy Berkeley DB backup/restore utilities to utilbin. Output diagnosis messages for Berkeley DB spooling in log_level log_debug. Bugfix: Fixed aimk -spool-* Review: SG MD-2004-04-21-1: Bugfix: - if SGE_ROOT was already set, the scripts throws thsi away and asked again - addapted testsuite to inst_sge Changed: testsuite, inst_sge Issue: 973 Review: AS (pending) MD-2004-04-21-0: Bugfix: - library path was wrong - cleanups removed SGEEE's - removed unused entries in config template Enhancem.: - basic backup/restore function - rc script checks, if master has started - start schedd, after comm to master is available - new entry in configuration Changed: inst_sge, install_modules Issue: 972 Review: CR, AS (pending) AA-2004-04-20-3: Bugfix: dlopen for all archs but hp11 for libcrypto openssl 0.9.7d Changed: aimk, aimk.site, distinst.site, all bins Issue: 914 Review: SG-2004-04-20-1: Bugfix: - fixed the build for DARWIN - set a max limit for MAX_DYN_EC in qmaster - updated the profiling in the scheduler Changed: sge_schedd, sge_qmaster Review: CR SD-2004-04-20-0: Bugfix: Modified auto user behavior to be more intuitive Fixed qconf to display delete_time Updated documentation and man pages Changed: qmaster, qconf Issue: 959 CR-2004-04-20-0: Bugfix: - check for duplicate component entries before adding new connections in cl_commlib_open_connection() - additional error handling fixes when open a new connection Changed: commlib EB-2004-04-20-1: Bugfix: load formula of the scheduler Issue: 330 Changed: sge_schedd, sge_qmaster Review: SG JG-2004-04-20-2: Enhancem.: qmake: In dynamic allocation mode it is now possible to specify resource requests for individual make rules. Changed: qmake Review: SG AA-2004-04-20-2: Bugfix: fixed install script, typo, tmp dir Review: MD Changed: inst_qmaster.sh, inst_execd.sh JG-2004-04-20-1: Bugfix: fixed linking of binaries doing dynamic loading on LINUX architectures. Review: AA AA-2004-04-20-1: Bugfix: qmon fixes: - interactive jobs, - blanks in complex attributes - delete time editability - job del deleted all instead of selected jobs - hostgroup member add callback Changed: qmon Issue: 582, 729, 840, 938, 382, 279 Review: EB AS-2004-04-20-0: Bugfix: MPI PE description update SunHPCT reference update removed loose integration with HPCT moved tight integration scripts to mpi/SunHPCT5 minor update to mpi/SunHPCT5/README SG-2004-04-20-0: Bugfix: fixed the scheduler profiling Changed: scheduler Review: JG AA-2004-04-20-0: Bugfix: compiler problem fixes EB-2004-04-20-0: Bugfix: stale resource_unknown_list Issue: 942 Changed: sge_qmaster Review: JG JG-2004-04-20-0: Enhancem.: Make dynamic spooling the default build rule, statically linked spooling methods can be choosen in aimk with the switches -spool-berkeleydb and -spool-classic. Bugfix: Fixed dynamic loading for darwin platform, partially fixed build with shared sge libraries. See also issue 976. Review: AA AH-2004-04-19-0: Enhancem.: Retry when rresvport() fails with EAGAIN as recommended by Goeran Uddeborg. Review: GU, JG AS-2004-04-19-2 Bugfix: use /bin/csh for script path Changed: util/sge_log_tee Issue: 956 SG-2004-04-19-1: Bugfix: fixed another sec fault in qsub. Changed: qsub Review: DT, HP AS-2004-04-19-1: Bugfix: get correct error message of hostname resolving problem Changed: util/install_modules/inst_execd.sh AS-2004-04-19-1 Bugfix: removed afs requirement for Linux insserv method Changed: util/rctemplates/* AS-2004-04-19-0: Bugfix: remove references to Enterprise Edition mode Changed: scripts/mk_ascman, man pages CR-2004-04-19-1: Enhancem.: qping will now be installed via distinst and copied into $SGE_ROOT/bin/$arch/qping Changed: install scripts Review: MD SG-2004-04-19-0: Bugfix: qsub is working again. Review: DT Changed: qsub AA-2004-04-16-1: Bugfix: typecast for time values fix for broken build on darwin JG-2004-04-19-0: Bugfix: Added files missing in checkin JG-2004-04-15-0. CR-2004-04-19-0: Cleanup: Changed name of module gridping to qping Changed: gridping AA-2004-04-16-1: Bugfix: typecast for time values fix for broken build on darwin CR-2004-04-16-1: Bugfix: Some performance fixes when max. file descriptor limit is reached. HP-2004-04-16-0: Enhancem.: Added flag to enable or disable drmaa_transfer_files via the configuration. Issue: 638 Changed: drmaa, japi, configuration Review: AH AH-2004-04-16-0: Bugfix: -soft submit option broken Issue: 917 Review: TBD JG-2004-04-16-0: Bugfix: Fixed build process for shared libs on AIX43 and AIX51. Changed README for berkeleydb: Berkeley DB has to be built with --with-mutex=POSIX/pthreads on AIX43. Review: AA, AS CR-2004-04-16-0: Bugfix: - flush unsent messages when a connection is going down. ( Event clients will now correct unreqister at qmaster ) Changed: commlib CR-2004-04-15-1: Bugfix: - always increment auto client id pointer to ensure that event clients won't reconnect with a client id from a former connection Changed: commlib Review: DT AH-2004-04-15-0: Bugfix: Fix sequential resource reservation broken by AH-2004-04-14-1. AH-2004-04-15-0: Bugfix: Fix problem with counter used to track number of JAPI threads that rely upon JAPI background thread service. Review: HP JG-2004-04-15-0: Enhancem.: Allow the admin to choose the spooling method (berkeleydb or classic) at installation time, if SGE was build with dynamic spooling (aimk -spool-dynamic). Changed: install scripts, testsuite Review: MD CR-2004-04-15-0: Bugfix: - output on qmaster clients when qmaster is not reachable - qconf -ke triggers "set_last_heard_from() not suppored by commlib" logging ( Issue 833 ) - fixed qmaster exceeds max file descriptor limit problem ( Issue 687 ) - fixed problem with reregistering of execd's at qmaster daemon - put information about max. used file descriptors into qmaster message file at startup of qmaster - delivering jobs will now fail if max. unheard value is reached, not when 2 load reports wasn't received - added new communication lists: endpoint_list - better performance at automatic id dispatching - added some thread trigger calls for a faster connection establishment/disestablishment - added 2 parameter to connect message: o port number where client is reachable and o autoclose mode of client (when service runs out of file descirptors the service will close connections where this mode is enabled) - service can now use random port if service port is specified with 0 - fixed bug in some list modify function of communication lib: return error code was wrong on some access functions - starting qmon with multithreaded communication lib. Issue(s): 833, 687 Review: Pending AH-2004-04-14-1: Bugfix: Per host slot limitation broken Issues: 968 AH-2004-04-14-0: Bugfix: Repeated schedd logging for unavailable exec hosts Issues: 927 EB-2004-04-14-0: Bugfix: reschedule_unknown broken Issues: 961 Changed: sge_qmaster Review: SG (part of SG-2004-04-08-0) JG-2004-04-14-0: Enhancem.: Reintegrated and verified dynamic loading of a spooling method. Changed: sge_qmaster, spoolinit, spooldefaults Review: EB CR-2004-04-13-2: Enhancem.: updated man pages (communication lib changes) CR-2004-04-13-1: Enhancem.: Added new utilbin binary: gridping JG-2004-04-13-0: Enhancem.: When an execd reports a job as running, but the job is in state deleted, send a kill signal for the job immediately, instead of waiting for the next repeated signal delivery to occur. Issue: 866 Changed: sge_qmaster Review: AH CR-2004-04-13-0: Bugfix: execd logging every 3 minutes Changed: gdi Issue: 915 SG-2004-04-11-0: Bugfix: The xml-output in qstat and the qstat xml schema files were not up to date Changed: qstat, qstat schema Issue: 964 RC-2004-04-11-1: Bugfix: Fixed a minor typo in manpage Changed: sge_conf.5 RC-2004-04-11-0: Cleanup: Fixed the IA-64 Linux build Changed: aimk SG-2004-04-10-0: Bugfix: - fixed the qdel help output - fixed qdel -u "*". The action is denied, if a non manager wants to execute it. Issue: 863, 903 Review: pending Changed: qmaster, qdel SG-2004-04-09-3: Bugfix: The qmaster loged errors in case that a job in T state was deleted. Review: EB Issue: 954 Changed: qmaster SG-2004-04-09-2: Bugfix: If the qmaster is killed while jobs are deleted, timed_event_thread error messages are reported when the qmaster is starting up again. The order in which the threads are stared has been changed. Review: AH Changed: qmaster SG-2004-04-09-1: Bugfix: - added the event handler for reschedule jobs - fixed the TYPE_SIGNAL_RESEND_EVENT event Review: CR Changed: qmaster SG-2004-04-08-0: Bugfix: - timed event handler: TYPE_SIGNAL_RESEND_EVENT was missing - some jobs stayed in the system in delete status. This happend due to some SGE v SGEEE code. Review: EB, AH, JG Changed: qmaster JG-2004-04-08-0: Bugfix: Fixed initialization of profiling in scheduler. Enabling profiling in scheduler did also enable profiling in qmaster. Changed: sge_qmaster, sge_schedd Review: SG AH-2004-04-07-0: Bugfix: Use a default 'weight_waiting_time' to 0 (#959) Review: SG JG-2004-04-07-0: Enhancem.: Berkeley DB spooling: Write two databases: A configuration database and a job database. Changed: sge_qmaster, spoolinit, spooldefaults, dbstat Review: EB SG-2004-04-07-0: Bugfix: - The or operator for complex entries with the type RESTRING was broken. - a minor issue in event master. It might deliver the event envelop without events. Changed: scheduler Issue: 958 Review: AA JG-2004-04-06-1: Cleanup: Some cleanup in berkeley db spooling. EB-2004-04-06-0: Bugfix: Usage retrieval/logging problems with tight PE jobs Issue: 929 Changed: sge_execd Review: JG JG-2004-04-06-0: Enhancem.: write intermediate accounting records to reporting file containing daily usage of long running jobs. Changed: sge_qmaster Review: EB MD-2004-04-06-0: Bugfix: - Improper variable comparison - autoinstall does not work - changed logfile dir - startup scripts on aix - nr of slots for host during execd inst - changed usage - typo, log text Issue: 629, 847, 897, 934, 937, 947 Changed inst_sge, install_modules, create_settings.sh rc_templates SG-2004-04-06-2: Bugfix: Finished the implementation for consumables as queue load_thresholds. It is working for parallel jobs now. (SG-2004-03-31-0) Changed: scheduler Issue: 464 Review: AH SG-2004-04-06-1: Bugfix: fixed the handling of job_load_adjustments for parallel jobs. Changed: scheduler Issue: 957 Review: AH SG-2004-04-06-0: Bugfix: fixed the profiling output in the mirror interface. Changed: scheduler Review: AH EB-2004-04-06-0: Bugfix: qconf -Mprj bad error message qconf -sprj gives bogus error message Issue: 923, 926 Changed: qconf HP-2004-04-05-0: Bugfix: shepherd created world writeable file, prolog execution as non-job user is broken Issue: 931,912 Review: AS, Changed: shepherd MD-2004-04-05-0: Bugfix: qstat -r prints low level error messages Issue: 930 Changed: qstat, schedd, qmon Review: AH EB-2004-04-05-0: Bugfix: qconf -aattr deletes the new attribute if it allready exists Changed: sge_qmaster Review: SG JG-2004-04-02-0: Bugfix: Fixed a bug in the spooling of array tasks introduced by JG-2004-03-30-0 Changed: qmaster Review: SG SG-2004-04-02-0: Bugfix: removed circular dependency between lib cull and lib obj. The change results in a limitation of the lWrite***(...) output. For the cull_packL.h cull elements are only the ids printed and not the element names. Changed: qmaster, scheduler Review: JG AA-2004-04-02-0: Bugfix: qalter -- option Issue: 844 Changed: qsub, qalter, qhold, qrls, qsh Review: SG JG-2004-04-01-0: Bugfix: Changes to reporting_params, sharelog, only took effect after a qmaster restart. Changed: sge_qmaster Issue: 940 Review: CR SG-2004-04-01-0: Bugfix: The profiling it turned of, when the schedd params are set to none. Changed: scheduler Issue: 936 Review: AA JG-2004-03-31-0: Enhancem.: Optimized non unique hash algorithm. Reenabled non unique hash on JB_owner in scheduler. Changed: all (libcull) Issue: 512 Review: EB SG-2004-03-31-0: Bugfix: First part of bugfix 464. Only parallel jobs and the resource reservation do not work yet. The user can add a consumable as a load_threshold and the scheduler will enable or disable the queues based on that value. The new code is not active in the case, that no consumables as load_thresholds are used. Changed: scheduler, qmaster, qmon, qstat, qhost Review: JG, AH is pending Issue: 464 JG-2004-03-30-0: Bugfix: At startup of an array task, all running array tasks were spooled. The usage container for finished pe tasks was not spooled. Changed: sge_qmaster Review: EB AS-2004-03-29-0 Enhancem.: qmon_help.ad does not need to be sed'ed anymore before it is installed Changed: dist/qmon/qmon_help.ad MD-2004-03-26-0: Enhancem.: Renamed ARCH variable to SGE_ARCH Fetch ADMINUSER from bootstrap during execd install setting file perms during automode Changed: inst_sge, inst_template, rc_templates AD-2004-03-26-0: Cleanup: Did fix build for Mac OS X 10.3 Changed: common, qconf Review: AH JG-2004-03-25-0: Enhancem.: Enhanced test program for CULL hashing. EB-2004-03-25-0: Bugfix: qconf -mq rejects pe_list config that is accepted with -aq (Fix of EB-2004-03-16-0) Changed: qconf, sge_qmaster Issue: 906, 856 Review: SG AS-2004-03-25-0: Enhancem.: add SGE_BINARY_PATH variable to job environment Changed: sge_execd, submit.1 Issue: 919 Bugtraq: 5018757 Review: DT JG-2004-03-24-1: Bugfix: Disabled non unique hash on JB_owner in scheduler. MD-2004-03-24-0: Bugfix: message, that qmaster didn't start always appears, check for running qmaster a bit later. Changed: rc template JG-2004-03-24-0: Bugfix: HP11 binaries ignored SHLIB_PATH due to missing linker flag +s. AH-2004-03-24-0: Bugfix: Earliest reservation not always found for parallel jobs. Scheduler seg faults on jobs with soft requests. Issue: 899, 911 Review: EB EB-2004-03-24-0: Bugfix: type in queue_conf(5) improved logging Changed: queue_conf(5) MD-2004-03-23-1: Bugfix: changed back to old boot order supressing the message, if execd starts first Changed: rc template, inst_sge MD-2004-03-23-0: Bugfix: fixed rc scripts boot order rc script output if master didn't start Issue: 784, 876 Changed: inst_sge, rc template JG-2004-03-23-1: Bugfix: fixed "Read from uninitialized" errors. Review: CR JG-2004-03-23-0: Testsuite: fixed test system_tests/clients/qsub. RH-2004-03-22-0: Bugfix: test case for issue 799 added to testsuite Issues: 799 Review: CR DT-2004-3-22-0: Bugfix: o Added -shell y|n option to qsub and update man page o Made -shell n the default for DRMAA clients o Made yes and no acceptable arguments to qsub y|n options and update man page o Minor cleanup in clients and spooling Issues: 476, 658 Changed: shepherd, qsub, qsh, execd, sge_spool.c, sge_job.h Review: EB, AH AS-2004-03-22-1 Bugfix: mk_dist script AS-2004-03-22-0 Bugfix: "about" output fix in qmon AH-2004-03-19-0: Bugfix: Changed value_len of drmaa_get_next_*() from int to size_t. EB-2004-03-19-0 Bugfix: Cleanup for the next main release Still checks for $GRD_ROOT, "spoolinit" may fail silently Issues: 122, 892 Changed: all, qconf.5 AS-2004-03-10-0: CVS Created V60_beta1_OS_TAG ---------------------------------------------------- SG-2004-03-18-0: Bugfix: - "qdel 0" is not allowed. - "qdel 45-54" results in an error and does not delete job 45. Issue: 893 Changed: qdel, qmod, qalter AA-2004-03-18-0 Bugfix: - qmon spinbox on lx24-amd, lx22-alpha - cluster dialogue crash, - Modify button in User dialogue Issues: 816, 840, 747 Changed: qmon AS-2004-03-18-2: Bugfix: Product name for qmon help file qmon_help.ad AS-2004-03-18-1: Enhancem.: added check for "localhost" resolving to installation procedure. Hostname may not be an alias for "localhost" or have IP 127.0.x.x AS-2004-03-18-0: Bugfix: use 32bit binaries on HP-UX 64 bit kernel in distribution CR-2004-03-18-0: Cleanup: added parameter check in cl_commlib_receive_message() Changed: commlib EB-2004-03-17-2: Bugfix: cluster queue instances like all.q@hostname are not resolved Issue: 850 Changed: sge_qmaster JG-2003-03-17-1: Testsuite: Cleanup and workaround for issue 884 EB-2004-03-17-1: Bugfix: qstat shows queue in A state although suspend_threshold is disabled Issue: 832 Changed: qstat, qhost, qmon DT-2004-03-17-0: Bugfix: o qsub -sync y now survives the qmaster going down. o JAPI allows client to register an error message callback for error messages from the event client thread. o qsub prints message when the qmaster goes away and comes back. o Left over checkin from DRMAA Java binding o Left over checkin from Flatfile Spooling Changed: JAPI, qsub, DRMAA Issue: 826 Review: SG MD-2004-03-17-3 Bugfix: merged fixes from beta branch to maintrunc adapted testsuite to changed message parts Changed: inst_sge, testsuite CR-2004-03-17-1 Bugfix: fixed performance problem in communication library in multithreaded mode. Changed: commlib Review: SG MD-2004-03-17-2 Bugfix: fixed broken fs check on not solaris and linux archs Changed: inst_sge, distinst Issue: 846 Review: CR JG-2004-03-22-0: Testsuite: fixed test system_tests/clients/qdel. Cleanup: Checkin of some files missing from earlier checkins, some additional docs and comments. CR-2004-03-17-0: Bugfix: CSP install failed on irix architecture Issue: 851 Review: MD SG-2004-03-17-0: Enhancem.: - Limited the number of dynamic event clients with a new qmaster_params entry: max_dyn_ec=XXX. - some cleanup in the event master thread - ensured that event clients with a fixed id are run by a manager. Changed: qmaster, man-page Review: AA JG-2004-03-17-0: Bugfix: fixed race condition in sge_shepherd (qlogin_starter for qlogin/qrsh jobs). Changed: sge_shepherd Issue: 871 Review: CR Bugfix: qstat -f output cluster queue name instead of queue instance name. Changed: qstat Review: SG EB-2004-03-17-0: Bugfix: Host aliases not supported with -q wc_qinstance/-masterq wc_qinstance Issue: 819 Changed: sge_qmaster Review: AH MD-2004-03-17-1: Bugfix: fstype was broken on IRIX 6.5 Changed: fstype Review: SG MD-2004-03-17-0: Enhancem.: added a new util to utilbin fstype, to check for local or mounted filesystems -> needed for Issue 846 Changed: fstype, utilbin/Makefile Review: CR RH-2004-03-17-0: Bugfix: improvements for bug 799 Changed: all Issue: 799 Review: SG AA-2004-03-16-0: Bugfix: - default_duration sconf param added - unused var compile warnings Issue: 839 Changed: qmon Review: EB EB-2004-03-16-0: Bugfix: Hostgroup identifiers not verified in queue_conf(5) Issue: 856 Changed: sge_qmaster Review: SG JG-2004-03-16-0: Testsuite: Cleanup and fix of tests bugs/issuezilla/193, system_tests/clients/qalter, system_tests/clients/qrsh SG-2004-03-16-0: Enhancem.: added the output of long qnames in qstat (again) Changed: qstat, qstat man page Review: JG EB-2004-03-16-0: Bugfix: mail bomb upon abort with tightly int. par. job Issue: 637 Bugtraq: 5020153 Changed: sge_execd Review: JG RH-2004-03-16-0: Bugfix: host group names has to start with a '@' Changed: qconf, qmaster Issue: 719, 740 Review: SG JG-2004-03-15-0: Testsuite: Fixed function changing queue attributes. DT-2004-03-15-0: Bugfix: o Fixed qconf line continuation in cases of long, unbreakable lines. o Fixed qconf parsing of trailing whitespace. o Moved DRMAA Java binding into more appropriate packages. o Added SISSL header to spooling parser. Changed: Flatfile spooling, DRMAA Issue: 818, 836 Review: JG RH-2004-03-15-0: Bugfix: Negative numbers for step in range parsing are not allowed. Changed: all Issue: 799 Review: SG AH-2004-03-15-1: Bugfix: Fix resource reservation problem for sequential jobs AH-2004-03-15-0: Doc: Made some fixes to calendar_conf(5) to make it's behaviour more clear. CR-2004-03-15-0: Testsuite: Adjustments for new GE install script output AD-2004-03-12-1: Bugfix: Did fix quote error. Changed: inst_execd.sh Review: AS AD-2004-03-12-0: Bugfix: Did fix broken symlinks for DARWIN. Changed: distinst Review: AS MD-2004-03-12-1: Bugfix: - added clear at the end of installation - local fs check for irix65 was broken Chaged: inst_sge Review: AD MD-2004-03-12-0: Enhancm: - added a autoinstall readme Changed: README-Autoinstall.txt SG-2004-03-12-2: Bugfix: execd died, when a load sensor was restarted. Issue: 835 bugtraq: 5019497 Changed: execd Review: AS AH-2004-03-12-1: Bugfix: fix access to free'd memory that caused schedd fp-exception under tru64 Review: AS CR-2004-03-12-0: Enhancm. Added a script to check install script syntax of infotext calls: Added: source/scripts/infotext_check_script_syntax.tcl Modified: source/scripts/infotext_msg_parse.tcl SG-2004-03-12-1: Bugfix: wrong qstat -F output with no requestable attributes Issue: 837 Review: EB MD-2004-03-12-0: Bugfix: wrong default duration in schedd configuration Issue: 838 Changed: inst_sge Review: SG JG-2004-03-12-2: Testsuite: Testsuite installation was broken due to changed messages. JG-2004-03-12-1: Bugfix: qconf -Aconf crashed. Issue: 834 Changed: all (libcull) Review: SG SG-2004-03-12-0: Bugfix: fixed problem with fix for issue 831 Issue: 831 Changed: qmaster Review: EB JG-2004-03-11-0: Enhancem.: Additional information about the build process of the Berkeley DB (architecture dependend configure options). AH-2004-03-11-2: Enhancem.: Fix problem of frequent "cannot resend unknown job" messages. Review: AD Issue: 822 AS-2004-03-11-1: Bugfix: installation fixes and formatting AS-2004-03-11-0: Enhancem.: Merged HPCT changes from 5.3 JG-2004-03-11-0: Testsuite: fixed tests functional/binary_submission and functional/migration. CR-2004-03-11-0: Bugfix: fixed error compiling qmon for aix43 Changed: qmon MD-2004-03-11-1: Bugfix: set default "y" for startup script inst Changed: inst_sge Review: AS SG-2004-03-11-0: Enhancem.: minor cleanup in lCopyElem in cull Changed: everything Review: JG MD-2004-03-11-0: Bugfix: - check for local filesystem on lx24-amd64 was broken - infotext argument parsing (irix65) Changed: inst_sge, infotext Review: CR AH-2004-03-11-1: Enhancem.: Fix sge_gethostbyname()/sge_gethostbyaddr() under tru64. Minor cleanup in new commlib. Review: AD AH-2004-03-11-0: Enhancem.: Allow inst_berkeley.sh to continue even without test regarding local file system for berkley db directory. Review: MD EB-2004-03-11-0: Bugfix: removing a suspend threshold doesn't unsuspend jobs Issue: 831 Changed: sge_qmaster Review: SG AS-2004-03-10-0: CVS Created V60_beta1_target_{BRANCH,TAG} ----------------------------------------------------------------- AH-2004-03-10-0: Enhancem.: Use more intuitive defaults for weight_ticket, weight_urgency and weight_priority. Lower impact of waiting time in urgency formula. EB-2004-03-10-2: Bugfix: calendar handling is broken Issue: 825 Changed: sge_qmaster Review: AH MD-2004-03-10-0: Bugfix: really bad typo Issue: 793 Changed: inst_sge JG-2004-03-10-2: Testsuite: Some fixes for the testsuite queue handling and tests bugs/issuezilla/193 and functional/migration. JG-2004-03-10-1: Enhancem.: Added datastructures for the reporting of long running jobs. Added an optimization for JG-2004-03-10-0. Changed: all EB-2004-03-10-1: Bugfix: Unsuspend on subordinate is broken Issue: 807 Changed: sge_qmaster Review: AH SG-2004-03-10-1: Bugfix: the japi event client does get qmaster goes down events and client shut down events. Part of 788 Changed: qmaster SG-2004-03-10-0: Bugfix: scheduler crashed during functional ticket computation on AMD64 - linux Changed: scheduler Review: CR Issue: 820 bugtraq: 5019584 EB-2004-03-10-0: Bugfix: qconf -mq rejects valid configuration Issue: 815 Changed: qconf Review: SG JG-2004-03-10-0: Bugfix: Signalling a slave task of a parallel job did not signal job. Issue: 810 Changed: sge_qmaster Review: EB MD-2004-03-09-0: Bugfix: install_execd ignores ignore_fqdn in bootstrap(5) Issue: 805 Changed: inst_sge AH-2004-03-09-1: Bugfix: Fixed qsub -w v to print more than one single reason. Fix problem with qstat -j scheduling info being truncated. Issues: 811, 812 Review: EB JG-2004-03-09-2: Bugfix: path of output file for pe start proc was incorrect. Issue: 808 Bugfix: fixed buffer overflow in shepherd. Bugfix: fixed uninitialized variable in commlib. Review: CR AH-2004-03-09-0: Bugfix: Reverted problematic do_ntix optimization done in AH-2004-03-08-0. Fix broken -masterq submit option. Issues: 809 Review: JG JG-2004-03-09-1: Testsuite: Adapted a number of changed messages and changed qstat output format. Fixed test functional/usage. JG-2004-03-09-0: Bugfix: Fixed memory access error in the deletion of automatic users. Bugfix: Writing of a job log entry in case of job error state crashed qmaster. Review: SG SG-2004-03-09-0: Bugfix: fixed a potential memory leak in the functional ticket computation. Changed: scheduler Review: EB AS-2004-03-09-1: Bugfix: reformat of qstat/qhost header lined for longer arch names AS-2004-03-09-0: Enhancem.: productization DT-2004-03-09-0: Enhancem.: o Finished defining native methods for JNI calls in DRMAA Java binding o Added SISSL header to all DRMAA Java binding files. Changed: DRMAA DT-2004-03-08-1: Bugfix: o Fixed the other half of logic hole in japi_wait. Changed: JAPI Issue: 802, 803 CR-2004-03-08-1: Bugfix: fixed bug in initializing of SGE_EVENT buffer for qmon Changed: qmon (qmon_init.c) CR-2004-03-08-0: Bugfix: setting default acknowledge timeout from 15 to 60 seconds. Changed: commlib DT-2004-03-08-0: Bugfix: o Fixed logic hole in japi_wait(). Changed: JAPI Issue: 798 AD-2004-03-08-0: Bugfix: Some qmaster threads did use the temporary message file. Changed: uti Review: JG AH-2004-03-08-0: Optim.: Skip computation of ntix/nurg/nprio if corresponding weights are 0.0 and report_pjob_tickets is set to 'false'. Assume fix range -1023 to 1024 when determining nprio. Changed: scheduler Review: EB MD-2004-03-08-0: Bugfix: qmaster crash in csp mode, if unsecure execd's are trying to connect to the qmaster Issue: 801 Changed: sec_lib, commlib Review: CR JG-2004-03-08-1: Bugfix: Added missing validation of database handle in spooling write function. Cleanup: Changed output level of a number of communication messages from INFO to DEBUG. Review: AD SG-2004-03-08-1: Bugfix: fixed memory leak in scheduler Changed: scheduler Review: AH SG-2004-03-08-0: Bugfix: - removed limit for flush_submit and flush_finish time - removed debug output from the scheduler - removed "schedd_params" from the cluster configuration Changed: install, scheduler Review: MD, AH EB-2004-03-08-1: Bugfix: missing newline in qstat -j output Changed: qstat EB-2004-03-08-0: Bugfix: qstat -g c takes to long time if there are many jobs in the cluster Changed: qstat Issue: 789 Review: SG DT-2004-03-08-0: Bugfix: Left over check-in from Friday o Fixed message for qconf -ks o Cleaned up some DRMAA output messages Changed: evm, GDI, DMRAA, answer, qmaster Issue: 785 JG-2004-03-08-0: Bugfix: Fixed dependencies for sge_sub_object.c Issue: 781 Review: AS DT-2004-03-07-0: Enhancem.: o Added DRMAA Java binding. Not yet complete. Changed: DRMAA SG-2004-03-07-0: Bugfix: - fixed the build for glinux - fixed the scheduler high profile - fixed the help output from inst_sge Changed: qstat, inst_sge DT-2004-03-05-5: Bugfix: o Changed error messages from DT-2004-03-05-4 to start with lower case letters. Changed: GDI Issue: 783 DT-2004-03-05-4: Bugfix: o Moved internal details from DPRINTF back into comm lib error message. Now there are two different messages. One is for the usual case of just not being able to connect, and the other is for all other cases. Changed: qmon, GDI Issue: 783 Review: pending DT-2004-03-05-3: Bugfix: o Moved internal details from comm lib error message into DPRINTF in sge_send_any_request(). Changed: qmon, GDI Issue: 783 Review: pending AH-2004-03-04-0: Bugfix: Job-queue matching in qstat broken Issue: 786 Review: AS AS-2004-03-05-0: Bugfix: wrong dependencies for sge_sub_object.c are now corrected through new script scripts/fix_dependencies Issue: 781 DT-2004-03-05-2: Bugfix: o Removed help output on qstat with no args Changed: qstat Issue: 782 CR-2004-03-05-1: Bugfix: sge_shadowd fails to create communication handle on takeover and exits. Issue: 776 Review: MD DT-2004-03-05-1: Bugfix: o Changed list separator from ',' to ' ' for hostgroups in qconf Changed: qconf Issue: 780 DT-2004-03-05-0: Bugfix: o Added missing error message for missing SGE_ROOT in qmaster o Removed duplicate comm lib "can't connect" error messages o Cleaned up qsub error messages o Added help output for qconf, qstat, qmod, and qalter with no args o Converted sge_send_any_request to use answer lists Changed: GDI, qsub, qconf, qalter, qmod, qstat, qmaster, evc Issue: 774 SG-2004-03-05-1: Bugfix: fixed a bug in changing build-in centries from strings to an other string type Issue: 759 Changed: qmaster SG-2004-03-05-0: Enhancem.: - updated the qstat schema files. - fixed a xml string encoding problem Changed: qstat schemas, qstat CR-2004-03-05-0: Bugfix: fixed broken CSP install Changed: install scripts, CULL feature list, feature functions Review: SG, MD Issues: 778, 779 DT-2004-03-04-3: Bugfix: o qconf now validates scheduler configuration settings Changed: qconf AH-2004-03-04-0: Bugfix: qselect/qstat -l selection wrongly considers load and utilization Issue: 771 Review: AS CR-2004-03-04-0: Bugfix: Rolled back change from CR-2004-03-02-2: all clients will now log error and critical messages to stderr again. Submitted cleanup issue: #744 Changed: all qXXXX clients MD-2004-03-04-1: Bugfix: - Pattern matching for -l q= doesn't work per default Issue: 759 Review: SG Changed: install DT-2004-03-04-2: Bugfix: o Removed duplicate error messages for missing SGE_ROOT in execd and schedd o Added missing error message for missing SGE_ROOT in qmon Changed: execd, schedd, setup_gdi.c Issue: 774 DT-2004-03-04-1: Bugfix: o Removed duplicate error message for missing SGE_ROOT Changed: age_arch.c Issue: 768 MD-2004-03-04-0: Bugfix: - Removed warning message at execd install - Formatting, text cleaning, typo Issue: 763 Changed: testsuite, inst_sge EB-2004-03-04-0: Bugfix: - misleading message of "qstat -explain a" if execd is down - slots in queue_conf(5) accepts 1,myslots=5, \ [@allhosts=4] Issue: 765, 755 Changed: qstat, qmon, qmaster Review: SG SG-2004-03-04-1: Bugfix: - another qstat -xml fix - fixed the build under glinux Changed: qstat, qconf SG-2004-03-04-0: Bugfix: was to quick checking in the qstat -xml fixes. When an element has no attribute it would print "(NULL)" instead of nothing. Changed: qstat Review: pending DT-2004-03-04-0: Bugfix: o Reenabled GUI logging in qconf in order to have flatfile spooling error messages printed. Changed: qconf Review: pending Issue: 768 AA-2004-03-03-1: Bugfix: - qi editability Isssue: 727 Changed: qmon Review: Pending SG-2004-03-03-5: Bugfix: fixed error message for wrong security mode string. Changed: all SG-2004-03-03-4: Bugfix: fixed several bugs in the qstat -xml output. The update for the XML schema file is still pending. Changed: qstat Review: pending SG-2004-03-03-3: Bugfix: The clients died when a wrong security mode is specified in the bootstrap file Changed: all Review: pending SG-2004-03-03-2: Bugfix: the product mode is displayed again with -help in the clientes Changed: all Review: pending AS-2004-03-03-1: Cleanup: SGEEE reference removals. Set version string to "6.0Beta" (code is not in Beta yet!) CR-2004-03-03-1: Cleanup: removed most of #ifndef ENABLE_NGC code. Still 52 ENABLE_NGC defines in source code. Review: pending Changed: removed unused code DT-2004-03-03-2: Bugfix: o Minor fix to DT-2004-03-03-0 Changed: Flatfile spooling DT-2004-03-03-1: Bugfix: o AS-2004-03-03-0 wasn't the best fix. Changed: Flatfile spooling AS-2004-03-03-0: Bugfix: Fix DT-2004-03-03-0 did not compile on all machines DT-2004-03-03-0: Bugfix: o Flatfile spooling now checks for duplicate attributes. Changed: Flatfile spooling Issue: 732 SG-2004-03-03-1: Bugfix: third part of removing SGE - simplefying the feature set code Changed: qmaster Review: pending AA-2004-03-03-0: Bugfix: - qi display - uppercase shortcuts - q customize - job dialogue segv for jobcustomize - ticket dialogue - qmon man page Issues: 734, 723, 680, 681, 725, 736, 747, 726 Changed: qmon Review: pending CR-2004-03-03-0: Bugfix: - additional parameter for cl_com_gethostname(), cl_com_cached_gethostbyname(), cl_com_cached_gethostbyaddr(),sge_gethostbyname() and sge_gethostbyaddr() which returns system error. - new function cl_com_get_h_error_string() which converts system error into a string. - cl_com_cached_gethostbyaddr() returns new error: CL_RETVAL_GETHOSTADDR_ERROR Issue: 674 Changed: commlib, gethostbyname, gethostbyaddr, gethostname Review: pending EB-2004-03-03-0: Bugfix: - job will be shown twice in qstat/qmon after restart of sge_qmaster - dbstat utility is unable to show certain objects - Show reason for queue and job error state in qstat Issue: 160, 757, 761 Changed: dbstat, sge_qmaster Review: SG NOTE: Cull data structure for ja_task has been changed. REEINSTALLION NECESSARY! SG-2004-03-03-0: Bugfix: - the clients print an error message in case of a bad bootstrap file (part 2). Changed: all Review: pending SG-2004-03-02-0: Bugfix: - fixed build for glinux - the clients print an error message in case of a bad bootstrap file (Part 1). A better error message is coming. Changed: all, scheduler Review: AH, EB DT-2004-03-02-2: Bugfix: o Fixed broken build o Added tmpdir to queue instance flatfile spooling Changed: Flatfile spooling, libsched Issue: 760 AH-2004-03-02-0: Enhancem.: Completed qstat -j schedd messages Fixed broken qsub -w e|w option Bugfix: Fixed schedd problem with jobs exceeding their prospecive duration for whatever reason. MD-2004-03-02-2 Bugfix: - adapted path to module files Changed: inst_sge MD-2004-03-02-1 Enhancem: - renamed installation files CR-2004-03-02-2 Bugfix: - removed some ENABLE_NGC code parts Cleanup: - qconf -mconf did not print out an error message when qmaster wasn't reachable - check_isalive() returns now commlib error - all clients will no longer log error and critical messages to stderr (log_state_set_log_gui(False)) (produces duplicate error messages) Issue: 662 Changed: all clients DT-2004-03-02-1 Bugfix: o Ulong parsing now complains about negative numbers o Tweaked freefix.pl search parameters NOTE: I have to check this in before I can compile or test it... Changed: o sge_object.c, freefix.pl Issue: 745 (not done yet) DT-2004-03-02-0 Bugfix: o Fixed qconf -mq help output. o Added -commit option to freefix.pl to overwrite original files. Changed: msg_common.h, freefix.pl Issue: 756 MD-2004-03-02-0 Bugfix: - added -dl option in qalter help output - changed product_mode to security_mode in bootstrap - fixed qmaster crash after getting wrong schedd config - changed weight values in preconfigured schedd setups Review: SG Changed: inst_sge, qalter, schedd CR-2004-03-02-1 Bugfix: - client will close it's own connection after closing connection timeout. Before this fix only the server closed the connection. - setting read, write and ack timeout (on tcp layer) back to 15 seconds. - cl_get_error_text() now return real error messages. Changed: commlib AS-2004-03-01-2 Enhancem.: make all test binaries for commlib and japi by default - needed for testsuite AS-2004-03-01-1 Enhancem.: renamed some install files and directories - moved startup template scripts to util/rctemplates - renamed startup template scripts to xxx_template where "xxx" is the same name as the final name - renamed 'inst_sgeee_modules' to 'install_modules' DT-2004-03-01-0 Enhancem.: o Updated qconf man pages for backslash line continuation. o Created script to fix lFree* assignment errors. AS-2004-03-01-0 Enhancem.: updated build README's AH-2004-01-03-0 Bugfix: Fix qstat bug with PEs using agv in urgency slots Issue: 710 MD-2004-01-03-1 Bugfix: - DB_CONFIG file copy as adminuser Changed: inst_sgeee MD-2004-01-03-0 Enhancem: - Add DB_CONFIG file (Berkeley Config) - Added new entry to schedd profiles Changed: inst_sgeee AA-2004-02-27-1 Bugfix: - q filtering -> segv Issue: 725 Changed: qmon AA-2004-02-27-0 Bugfix: - qi dialogue shows only fixed number of queues - Modify button in user configuration - cq modify queue name line insensitive - time helper dialogue - comment, msg corrections Issue: 747, 739, 736, 723 Changed: qmon, msg header changed -> all is recompiled AS-2004-02-27-2" Enhancem.: added "-nosource" switch" to scripts/distinst AS-2004-02-27-0 Enhancem.: changes to the install procedure Changed: scripts/distinst and related minor files Renamed: inst_sgeee -> inst_sge CR-2004-02-27-4: Testsuite: compile_source procedure now supports new "myinst" behaviour. Changed: testsuite CR-2004-02-27-3: Cleanup: setting larger default timeouts for communications Changed: commlib CR-2004-02-27-2: Bugfix: changed DEBUG_FLAG from -g3 to -g2 for ALPHA5 (tru64) sytems (-g3 produced segmentation faults in gethostname binary) Changed: aimk SD-2004-02-27-0: Bugfix: Fixed checking of read return code Changed: execd Issue: 713 DT-2004-02-27-0: Bugfix: o Rearranged japi_exit() to prevent the reopening of the close comm lib connection. o qconf -sq @... now returns a friendly error message. o qconf now requires all fields to be present for all options. o Fixed various qconf bugs Changed: Flatfile Spooling, JAPI, qconf Issue: 692, 720 SG-2004-02-27-0: Bugfix: increased the timeout in qdel, since it might take a bit longer to delete the jobs (with array tasks). Changed: qdel Review: CR CR-2004-02-27-1: Bugfix: - ignore_fqdn and default_domain parameters wasn't used in communication lib. Changed: commlib AS-2004-02-27-0: Bugfix: corrected wrong exit status from tests" Changed: aimk CR-2004-02-27-0: Bugfix: - Use thread signal handler for qsub only when -sync yes or -now yes is used (when a thread is started) otherwise the default sge signal handler is used. - japi_init_mt(): block PIPE signal before doing sge_gdi_setup() Changed: qsub, japi AS-2004-02-26-0: Enhancem: - Tru64 compile port - QIDL remove, header file cleanup - Makefile cleanup (did not fix not working rules) - Makefile dependency correction - Pass same flags to qmon make call as for core system in aimk NOTE: Call "aimk clean; aimk depend" after checkout DT-2004-02-26-0: Bugfix: o qsub doesn't hang anymore. o JAPI no longer uses signals to stop the event client thread. o japi_exit() no longer blocks when another is already in japi_exit(). o Fixed various JAPI bugs. o Simplified qsub's logic. Changed: qsub, JAPI, comm lib (minor) Issue: 683 Review: pending SG-2004-02-26-1: Bugfix: qmaster was not able to read in the scheduler configuration in classic spooling mode. Issue: 717 Changed: qmaster EB-2004-02-26-1: Bugfix: qmon shows invalid messages if "show detached settings" button is pressed renamed "qconf -sick" to "qconf -sds" Changed: qmon Issue: 707 Review: AA CR-2004-02-26-1: Bugfix: using cl_com_ignore_timeouts() in default signal handler to induce communication functions to return without timeout in order to terminate sge client programs immediately after sending TERM signal. Changed: all binaries Issue: 686 ( fixes all client programs expect "qsub" ) AH-2004-02-26-0: Doc: Mention sched_conf(5) monitor=1 params setting. Document improved handling of wildcard PE request. CR-2004-02-26-0: Bugfix: don't sleep when qevent can't open connection to qmaster and signal TERM is sent Changed: qevent SG-2004-02-26-0: Enhancem: added halfelife_decay_list to the scheduler config man page. AA-2004-02-26-0: Bugfix: cplx dialogue kept highlighting after closing Changed: qmon EB-2004-02-26-0: Bugfix: qstat -explain did not show messages qalter -q is not rejected. (reintroduced by AA-2004-02-21-0) Changed: qstat, sge_qmaster Issue: 705, 688 Review: SG SG-2004-02-25-1: Bugfix: fixed the time bases for YY specification. It referes to to the range of 1970 till 2069. Chaged: qacct, qsub, qalter,... Issue: 709 bugtraq: 5019595 SD-2004-02-25-0: Bugfix: Fixed file descriptor leak which caused execd to run out of file descriptors Changed: execd Issue: 713 AH-2004-02-25-0: Enhanc.: Added 'default_duration' to sched_conf(5). Doc: Man pages changes for reservation scheduling. Bugfix: Wrong error message for qsub -R Issue: 694 Review: AD DT-2004-02-25-0: Bugfix: o Fixed qconf -sq queue@hostname output. o Fixed DRMAA example program Issue: 701, 704 Changed: DRMAA, qconf SG-2004-02-25-0: Bugfix: extended the scheduler check for invalid configurations. Issue: 695 Changed: scheduler SD-2004-02-24-0: Cleanup: man page updates, automatic user documentation Issue: 636 AD-2004-02-24-1: Bugfix: Did fix NULL pointer assignment. Changed: qstat Issue: 696 MD-2004-02-24-0: Bugfix: changed messages Issue: 676 Changed: inst_sgeee AD-2004-02-24-0: Bugfix: Did fix CULL run time type error. Changed: qmaster Issue: 699 AH-2004-02-23-2: Bugfix: Improved diagnosis output of drmaa_init() in case $SGE_ROOT is not set AH-2004-02-23-1: Bugfix: Added ":" to DRMAA example.c output path. Improved example.c diagnosis capabilities. AD-2004-02-23-0: Bugfix: Did add missing calendar event handler. Changed: qmaster AA-2004-02-23-0: Enhancem.: qmon CQ dialog, sensitivity toggling, lock icons Changed: qmon AH-2004-02-23-0: Bugfix: Fix broken reservation for parallel jobs. Refresh CE_doubleval in JB_hard_resource_list after qmaster restart. Fix broken category optimization. Fix problem with handling default requests. Issue: 700 AA-2004-02-21-0: Enhancem.: qmon CQ changes and fixes partial fix of bool, lBool casting RR dialogue changes fixing of broken functionalities (Save settings) Changed: all due to lib dependencies JG-2004-02-20-1: Enhancem.: Added some documenation related to reporting, cleanup of man page bootstrap.5. Testsuite: Fixed some problems with tests functional/accounting and bugs/issuezilla/440. AS-2004-02-20-1: Cleanup aimk cleanup, makefile cleanup, porting Review: AA (partial review and help) MD-2004-02-20-1: Bugfix: - fixed wrong schedd default configuration Changed: schedd templatefiles (installation) CR-2004-02-20-0: Bugfix: reaching max possible connection count will disable accepting/creating new connections. This is the part of fixing Issue #687 Changed: communication lib AD-2004-02-20-1: Cleanup: Did remove obsolete code. Changed: shadowd AH-2004-02-20-0: Bugfix: Made libdrmaa.so again a self-contained library that requires no other Grid Engine libraries to be linked against DRMAA application. Added DRMAA example to default build rule. AD-2004-02-20-0: Bugfix: Generate error message if qmaster is already running. Changed: qmaster, shadowd, daemons/common Review: CR Issue: 684 EB-2004-02-20-0: Bugfix: Added missing documentation Issue: 691 SG-2004-02-20-0: Bugfix: - evaluated 53patches - fixed test 540 for the testsuite Changed: testsuite Review: CR EB-2004-02-20-0: Bugfix: Soft queue requests are not rejected if queues are not requestable. qalter -q is not rejected. Changed: sge_qmaster Issue: 688, 689 Review: SG MD-2004-02-20-0: Bugfix: - new dialog messages Changed inst_sgeee, testsuite Issue: 676 JG-2004-02-20-0: Enhancem.: Tune locking parameters of Berkeley DB. MD-2004-02-19-0: Bugfix: - wrong perms and owner for bootstrap file - typos Changed: inst_sgeee, testsuite Issue: 677 JG-2004-02-19-2: Testsuite: Fixed testsuite installation: Use SGE_SINGLE_LINE as introduced by DT-2004-02-19-1 to avoid parsing problems due to splitted long lines. AD-2004-02-19-2: Cleanup: Did remove '-lj' command line argument. Changed: qmaster, common, man SG-2004-02-19-3: Bugfix: fixed qstat -s z -r Changed: qstat EB-2004-02-19-1: Bugfix: It was possible to delete used cluster queues Changed: sge_qmaster Review: SG Issue: 679 DT-2004-02-19-1: Bugfix: Did various minor qconf & FF bug fixes. qconf now supresses line continuation when the SGE_SINGLE_LINE env var is set. Added new test program to test compatability between classic and flatfile spooling. Changed: qconf, flatfile spooling HP-2004-02-19-1: Bugfix: - qsub -S :/ crashed sge_qmaster Issue: 682 HP-2004-02-19-0: Enhance: - filestaging for drmaa - prolog handles application error Changed: all Review: AH Issue: 638, 377 AD-2004-02-19-1: Cleanup: Did remove '-s' command line argument. Changed: qmaster, common, man DT-2004-02-19-0 Bugfix: Missing files from JGrid package Changed: JGrid howto AS-2004-02-19-1 Cleanup: Crude port to HP1164. Just COMPILES!!!! No testing at all done yet! "qtcsh" is not really working. Reason is that "-ltermcap" does not exist on our system as 64-bit library. Use "-lcurses" instead. However tcsh prints at startup: SG-2004-02-19-2: Bugfix: - impemented qhold and qrls in qalter and removed the qhold.sh script. - corrected the qresub help output Changed: qhold, qrls, qresub, qalter, distinst Review: MD Issue: 675 AD-2004-02-19-0: Cleanup: Did remove '-nostart-commd' command line argument. Changed: qmaster, common, man AH-2004-02-19-1: Bugfix: sge_schedd crashed when parallel jobs were assigned into more than one queues instance of the same host. AS-2004-02-19-0 Cleanup: removed/add compiler warning level switches compile with +O2 on HP11 due to compiler bug Unify HP11 and HP1164 settings in one section Remove Solaris x86 suppress warning settings AH-2004-02-19-0: Bugfix: Jobs were not scheduled into second queue instance of a host Review: JG Issue: 672 SG-2004-02-19-1: Bugfix: - scheduler did not get the global configuration - event master worked outside an array. Changed: scheduler, qmaster Review: CR Issue: 678 EB-2004-02-19-0: Bugfix: qsub -l q= was not rejected Issue: 671 Changed: sge_qmaster Review: JG JG-2004-02-19-1: Testsuite: Testsuite install output warnings due to changed defaults for weight_* schedd config attributes. JG-2004-02-19-0: Bugfix: Added source for dbstat. SG-2004-02-19-0: Bugfix: the event should not accept new event clients when the master is going down. Changed: qmaster JG-2004-02-18-1: Bugfix: Fixed build for dbstat. JG-2004-02-18-0: Testsuite: Fixed testsuite installation - it failed as schedd conf attribute weight_jobclass no longer exists. Did some fixes to tests system_tests/clients/qconf/check.exp and system_tests/clients/qdel/check.exp MD-2004-02-18-0: Bugfix: - cleanup during execd uninst - if spooldir exists -> ask for deleting - testsuite adaption Changed: inst_sgeee, testsuite CR-2004-02-18-1: Bugfix: - qevent will not immediately try to re-register on event client timeout - 2nd started scheduler will exit - communication errors and event client timeout is are now handled when event client wants to get events ( function: ec_get() ) Changed: event clients ( sge_schedd, qevent, qsub, japi lib) Review: SG DT-2004-02-18-4: Bugfix: Revamped JAPI signal handling to be more POSIX friendly and less dangerous. Reworked JAPI event client thread so that it doesn't hang the process whenever something goes wrong. JAPI now shares one comm lib connection between the main thread and the event client thread. Changed: JAPI Review: pending AD-2004-02-18-1: Cleanup: Did remove *currently* dispensable mutex attributes. Did add some additional commentaries. Changed: qmaster SG-2004-01-18-3: Bugfix: - fixed qmaster crash in case of "qhold all" - removed -uall switch from qalter - removed "all" flag from qalter. "all" is a job name now. - adjusted the qhost help output Changed: qmaster, qalter, qhost Review: MD Issue: 670 DT-2004-02-18-3: Bugfix: Moved basic sharetree functionality out of libsched into libsgeobj to try to eliminate odd dependencies. Still a work in progress. Changed: sge_support, sge_sharetree Review: JG DT-2004-02-18-2: Bugfix: Fixed JGrid Howto mistakes. Changed: JGrid Howto SG-2004-02-18-2: Bugfix: - the qmaster did not terminate, when no event client was registered. - the qmaster shutdown event is delivered to all clients Changed: master Review: CR, AD Issue: 669 CR-2004-02-18-0: Enhancem.: - added error function callback for communication errors - connect response message reports now if connection endpoint is already connected with service endpoint Changed: communication lib DT-2004-02-18-1: Bugfix: Added missing -sconf param in qconf Changed: flatfile spooling SG-2004-02-18-1: Bugfix: the total update events were delivered with the next regular event. Changed: master SG-2004-02-18-0: Cleanup: first part of doing the event client matching in the event master thread. Changed: master Review: DT AD-2004-02-18-0: Bugfix: Did fix 'qconf -secl' and 'qconf -sss'. Changed: qmaster, evm Review: SG Issue: 666 DT-2004-02-18-0: Bugfix: Fixed backslash line continuation for all objects in qconf. Will break on spaces or commas. Propogated removal of SGE/SGEEE through qconf and flatfile spooling. Fixed build on Solaris x86. Fixed other minor qconf and flatfile spooling issues. Changed: qconf, qmon, flatfile spooling Review: JG AS-2004-02-17-0: Cleanup: aimk cleanup, #define cleanup, IRIX 6.5 compile Review: EB SD-2004-02-17-0: Cleanup: - Removed all jobclass references from qmon, scheduler, and scheduler configuration - Added display of users which map to the default user node to qmon Changed: qmon Issue: 636, 643 JG-2004-02-17-0: Bugfix: Mirror interface didn't propagate event client errors. Make qevent analyze return code from mirror interface. Review: CR AA-2004-02-17-0: Bugfix: Shell attribute was missing in Qconfig Segv when attribute could not be applied Changed: qmon Review: EB AH-2004-02-17-0: Bugfix: Fixed dispatching of parallel jobs with reservation disabled. Review: EB DT-2004-02-17-0: Bugfix: Fixed segv in qconf -msconf Changed: qconf Review: AH JG-2004-02-16-2: Testsuite: Checked and/or fixed following testsuite tests: functional/binary_submission functional/job_environment system_tests/clients/qalter JG_2004-02-16-1: Bugfix: Serialize (re)open actions of Berkeley DB by multiple threads. Enhancem.: Do spooling of hostgroups (comprising spooling of multiple queue instances) within a transaction. Review: AD, EB DT-2004-02-16-0: Bugfix: Fixed various qconf bugs. \ line continuation is temporarily broken. Changed: qconf AD-2004-02-16-0: Bugfix: 'qconf -ks' now does output message about scheduler shutdown. Changed: qmaster, evm Review: SG Issue: 665 SG-2004-02-16-1: Bugfix: - second part of removing SGE - fixed processing of scheduler configuration in scheduler - fixed event_master delivery time. Changed: all Review: CR MD-2004-02-16-0: Bugfix: Fixed some Messages, spooldir default Testsuite: Adapted to new messages Enhancem: Added 3 default scheduler setups Changed: inst_sgeee, testsuite Review: SG EB-2004-02-16-1: Bugfix: wrong queue selection in qmon and qstat Changed: qstat, qmon Review: AA CR-2004-02-16-1: Bugfix: Disabled reconfigure in qmake/IRIX64/glob/Makefile CR-2004-02-16-0: Testsuite: Fixed still existing problem: don't send job start command to a remote shell when a former job start is not yet executed. AA-2004-02-16-0: Bugfix: dependencies in Makefiles from ancient paths removed JG-2004-02-16-0: Testsuite: Checked and fixed test functional/usage. Source tcl_files/private.tcl to enable private installation settings. Make testsuite run with architecture lx2[46]_amd64. EB-2004-02-16-0: Bugfix: eliminated ambiguous message ids removed obsolete messages Changed: all AD-2004-02-16-0: Cleanup: Did remove superfluous debug output. Did adjust comments. Changed: evm SG-2004-02-15-0: Bugfix: the event thread works with a fixed interval to deliver events (every second) Changed: qmaster Review: pending Issue: 668 JG-2004-02-14-2: Bugfix: Fixed broken build on lx24-amd64. SG-2004-02-14-1: Bugfix: the event master ignored the flush settings. Every flush was handled as flush = 0 seconds. Changed: qmaster Review: pending JG-2004-02-14-1: Bugfix: Make spoolinit (creation of the spooling database) fail, if the database already exists. Review: pending JG-2004-02-14-0: Testsuite: Checked and/or fixed following testsuite tests: bugs/issuezilla/226 bugs/issuezilla/314 bugs/issuezilla/402 bugs/issuezilla/403 bugs/issuezilla/423 bugs/issuezilla/440 bugs/issuezilla/478 bugs/issuezilla/509 bugs/issuezilla/511 bugs/issuezilla/524 bugs/issuezilla/527 bugs/issuezilla/540 bugs/issuezilla/445 api/drmaa SG-2004-02-14-0: Bugfix: fixed the build for linux DT-2004-02-13-0: Enhancem: qconf now uses flatfile spooling instead of classic spooling. Might be broken for EH and PE because of the new RUE type. Not yet fully cleaned and commented. Bugfix: Fixed a variety of classic spooling, qconf, and other bugs. Changed: Flatfile spooling, classic spooling, qconf, sgeobj Review: JG CR-2004-02-13-1: Testsuite: throughput test: don't send job start command to a remote shell when a former job start is not yet executed. CR-2004-02-13-0: Testsuite: fixed bug when setting up user portlist in user configuration SG-2004-02-13-0: Cleanup: first part of removing sge from sgeee Changed: all Review: AA AD-2004-02-13-0: Cleanup: Did add some missing comments. Changed: evm JG-2004-02-13-1: Bugfix: fixed broken test system_tests/clients/qrsh. JG-2004-02-13-0: Enhancem.: Extended job log information. Improved README for Berkeley DB build. Bugfix: Fixed broken build for lx24-amd64. EB-2004-02-13-0: Bugfix: Fixed broken build AH-2004-02-12-1: Bugfix: Complex requestable field was not spooled. AH-2004-02-12-0: Enhancem: Initial revision of resource reservation. Man pages and devel spec to follow. Source comments to be completed. Currently broken - qstat -j schedd info messages - -w e|v|w submit(1) option Soft requests to be verified. Host/queue category optimization to be revived. Testsuite support to be added. Enhancem: Added CULL functions lAddSubList()/lAddDouble()/ lAddUlong() Bugfix: Fixed schedd crash with large array jobs. Fixed problem with whitespaces in normalized time specifications. Review: FF/EB AD-2004-02-12-0: Bugfix: Finish 'AD-2004-02-08-0'. Event Master (evm) is now a self-contained modul and does use its own thread to send events to event clients. Issue: 656 Changed: evm, qmaster CR-2004-02-11-4: Testsuite: - removed "press RETURN" in cleanup_system procedure of install_core_system test CR-2004-02-11-3: Cleanup: - removed unused static function qmaster_log_flush_func() CR-2004-02-11-2: Bugfix: - qmaster communication is now multithreaded again - qmaster can now bind port < 1024 as root (setup of communication is done earlier in startup phase of qmaster) Changed: qmaster Review: AD SG-2004-02-11-0: Bugfix: scheduler died, when the qmaster removed automicaly created users. Changed: qmaster Review: AA AA-2004-02-11-1: Enhancem: qmon - explain functionality - cq filtering - funcs from qstat relocated to be shared with qmon Review: SG Changed: qstat, qmon AA-2004-02-11-0: Bugfix: Compiler warnings removed Review: CR Changed: all CR-2004-02-11-1: Testsuite: Reworked throghput performance test CR-2004-02-11-0: Cleanup: Fixed message macro name in qsub JG-2004-02-11-1: Cleanup: Cleanup and enhancement of error messages in case of spooling errors. Review: pending DT-2004-02-11-0: Bugfix: Fixed aimk -help output. Made the -secure transition more user friendly. Changed: aimk CR-2004-02-11-0: Testsuite: Testsuite will not use -secure switch for aimk. This switch is not supported by aimk anymore. JG-2004-02-11-0: Testsuite: performance/throughput was broken for SGE 5.3. Fixed bugs/issuezilla/126, bugs/issuezilla/140, bugs/issuezilla/141, bugs/issuezilla/171, bugs/issuezilla/193 SD-2004-02-10-1: Enhancem: Added automatic user fields to qmon cluster configuration Changed: qmon Issue: 636 MD-2004-02-10-0: Bugfix: -fixed some typos -fixed local execd spool dir support -added local execd spool dir support (auto mode) -adapted testsuite -help text Changed: inst_sgeee testsuite SG-2004-02-10-0: Bugfix: fixed memory leak in scheduler Changed: scheduler Review: CR AD-2004-02-10-0: Enhancem: Did add wrapper functions for POSIX read-write locks. Changed: lck JG-2004-02-10-0: Testsuite: Fixed install re_init. Allow tests that are only valid for a certain Grid Engine version. Cleanup functional/access_lists. Fixed setup and cleanup of performance/throughput. Fixed set_queue for cluster queues. SD-2004-02-10-0: Bugfix: - Various qmon updates for issue #643 Changed: qmon Issue: 643 SD-2004-02-09-0: Enhancem.: - Added POSIX priority support to the SGEEE priority formula. - Added -pri qstat option. - Added weight_priority to scheduler configuration - Various qmon updates for issue #636 Changed: qmaster, schedd, qstat, qmon Issue: 636 CR-2004-02-09-2: Testuiste: Fixed problems with admin mail address in install core system test SG-2004-02-09-2: Enhancem.: added the schema for qstat SG-2004-02-09-1: Bugfix: -qstat -r -xml died. Changed: qstat AD-2004-02-09-0: Bugfix: - Did move 'sge_gdi_tsm()' out of evm. - evm no longer does link libgdi - evm no longer does link libngc and libngclists Issue: 656 Changed: qmaster, evm AS-2004-02-09-1: Enhancem.: porting changes for HP (begin native 64bit port), IRIX (support only IRIX 6.5) and Linux (AMD64 and IA64). No functional tests done yet. Only #define cleanup and changes JG-2004-02-09-0: Testsuite: Enable testsuite installation on aix51. CR-2004-02-09-1: Bugfix: Fixed compile problem for AMD64 AS-2004-02-09-1 Bugfix: pass -Wno-strict-prototypes to gcc on Linux to compile sec_lib.c without error (introduced -Wstrict-prototypes for all Linuxes with AS-2004-02-06-1) CR-2004-02-09-0: Testsuite: Fixed install_core_system test on hosts where no local spool directory is defined. AA-2004-02-09-0: Bugfix: wrong define in basis_types.h leads to execd crash Changed: all Review: SG SG-2004-02-09-0: Bugfix: - qstat -urg crashed - added "-urg" to the qstat -help output - qstat -ext -xml did not display ticket information Changed: qstat Review: AA AD-2004-02-08-0: Bugfix: Make over of the event master (evm) external interface. This is needed in preparation for a self-contained event master module. Issue: 656 Changed: qmaster, evm AS-2004-02-06-1 Enhancem.: define cleanup, IA64 and AMD64 port begin" AS-2004-02-06-0 Enhancem.: aimk cleanup Introduce define -DTARGET_32BIT or -DTARGET_64BIT use only one section for Linux (except SLINUX) SG-2004-02-06-1: Enhancem.: updated the complex man page. Changed: submit.1, complex.5 CR-2004-02-06-1: Bugfix: Fixed broken build for linux Changed: sec_lib.c CR-2004-02-06-0: Bugfix: Gridengine system with csp feature not working - modified csp feature for new communication Issue: 657 Changed: all binaries, install scripts, japi Review: AA AD-2004-02-06-1: Bugfix: 'sge_remote_event_delivery_handler' is now called after message. Changed: qmaster Review: CR AD-2004-02-06-0: Bugfix: Use lock 'LOCK_GLOBAL'. Changed: spool AD-2004-02-05-5: Cleanup: Did remove all locks except 'LOCK_GLOBAL'. Changed: lck AD-2004-02-05-4: Bugfix: Use lock 'LOCK_GLOBAL'. Changed: qmaster AD-2004-02-05-3: Bugfix: Use lock 'LOCK_GLOBAL'. Changed: qmaster AD-2004-02-05-2: Bugfix: Use lock 'LOCK_GLOBAL'. Changed: qmaster AD-2004-02-05-1: Bugfix: All extern functions now lock 'LOCK_GLOBAL'. Changed: evm EB-2004-02-05-0: Bugfix: Changed man page Changed: sge_types.1 MD-2004-02-05-0: Testsuite: - adapted testsuite to new installation script Bugfix: - some text - local execd spooldir support Changed: testsuite, inst_sgeee SD-2004-02-05-0: Enhancem.: Replaced use of qsub option -p with -js for specifying GEEE job shares to influence the GEEE share tree and functional policies. Changed: qmaster, schedd, qsub, qalter, qhost, qmon, qsh, qstat, submit(1), user(5) Issue: 636 AD-2004-02-05-0: Bugfix: If 'sge_mutex_lock/unlock()' do fail, the error code is now part of the log message. Changed: lck SG-2004-02-05-0: Enhancem.: updated man pages to reflect my latest changes Changed: qdel.1, qmod.1, qstat.1, sge_types.1, submit.1 JG-2004-02-05-0: Testsuite: Fixed functional/tight_integration. JG-2004-02-04-4: Testsuite: Fixed system_tests/clients/qmod/general, del_queue, cleanup of queue functions. SG-2004-03-04-1: Bugfix: the update list event has to be delivered allways. In case, that all elements got removed due to filtering, the event was not sent. This is changed. Changed: qmaster Review: DT SG-2004-02-04-0: Bugfix: - qdel "*" deleted only some jobs but not all. - the max deletion time in qmaster is set to 3 seconds. Changed: qdel, qmaster Review: DT AS-2004-02-04-0: Enhancem.: "aimk -32" creates Solaris 32 bit binaries on 64 bit kernel JG-2004-02-04-3: Testsuite: Fixed system_tests/clients/qrsh and functions enabling and disabling queues. AA-2004-02-04-0: Bugfix: fixed AIX51 spinbox compile problem JG-2004-02-04-2: Bugfix: Fixed build of rshd on Solaris 7 32 bit. Review: AS DT-2004-02-04-0: Bugfix: Fixed JAPI session initialization bug which caused qsub to segfault if exiting due to failed job submission. Changed: JAPI Review: SG JG-2004-02-04-1: Testsuite: fixed testsuite for SGE 5.3 JG-2004-02-04-0: Enhancem.: Added a utility to dump the list of objects in the Berkeley DB database, dump objects and delete objects. Review: pending SG-2004-02-03-1: Bugfix: qacct only worked on queues and queue instances, which are currently in the system. Removed ones were not accessable. This is changed. Changed: qacct Review: EB EB-2004-02-03-0: Bugfix: - It was possible to delete host groups which were referenced in cluster queue objects. - Due to fix AD-2004-02-03-0 tight integration for parallel jobs works correcly now. Changed: qmaster Review: SG JG-2004-02-03-0: Testsuite: Fixed test functional/access_lists. Enhancements and bugfixes in the handling of cluster queues. AD-2004-02-03-0: Bugfix: Did add event handler for 'TYPE_JOB_RESEND_EVENT' Changed: qmaster Review: EB MD-2004-02-03-0: Bugfix: - Text - fixed install_qmaster, install_execd Changed: - inst_sgeee SG-2004-02-03-0: Bugfix: - added the user id to the qdel error messages. - some code cleanup Changed: qmaster Review: JG CR-2004-02-03-1: Bugfix: Fixed build for tru64 (no prototype for "sge_get_execd_port()") CR-2004-02-03-0: Testsuite: HTML compile output links of compile output HTML file are now relative to HTML-directory. MD-2004-02-02-2: Bugfix: - Text Changed: - inst_sgeee MD-2004-02-02-1: Bugfix: - renamed sge_config to inst_sgeee - moved sge_config_modules to util/inst_sgeee_modules - adaptions of distinst Changed: - sge_config, distinst SG-2004-01-02-1: Bugfix: reverted the changes from SG-2004-02-01-0. Reopend the issue 476 again. AD-2004-02-02-0: Cleanup: qmaster threads now do call 'sge_qmaster_thread_ init()' at the beginning. MD-2004-02-02-0: Bugfix: - DB installation - Test for local fs, new archs - Text Changed: - sge_config CR-2004-02-02-0: Bugfix: Fixed problem "can't get communication handle" when application calls check_isalive() before a communication handle is created. Now all communication handle are created in prepare_enroll() Review: pending JG-2004-02-02-0: Enhancem.: The event client layer handles shutdown of qmaster (event QMASTER_GOES_DOWN). In this case, the event client will try to reregister itself at the (new) qmaster. Changed: each eventclient (sge_schedd, qevent) Review: SG AD-2004-02-02-0: Cleanup: Did add function to initialize a qmaster thread. Changed: qmaster CR-2004-02-02-0: Bugfix: - fixed csp install: check for CRYPTO_ strings in qmaster binary, not for sec_init to test if binary was compiled with -secure option MD-2004-01-31-1: Bugfix: - Shutdown of Qmaster with rcscripts under Linux - Test for local fs under Linux - Text Changed: - sge_config CR-2004-02-01-0: Enhancm.: Added missing functions: cl_commlib_get_last_message_time() (former last_heart_from() ) and check_isalive() Event client: recreate connection object after qmaster shutdown in ec_register() increased default timeout for connection lifetime Changed: all binaries DT-2004-02-01-0: Bugfix: Increased MAX_STRING_SIZE from 2048 to 8192 to allow for longer PATHs. Issue: 538 Changed: sge_basis.h SG-2004-02-01-0: Bugfix: in case a job start failed, the queue is set into error state when it is a job script but not for binary jobs. This has be fixed. The queue is set into error state, when ever a job start fails. Review: pending Issue: 476 Changed: execd JG-2004-02-01-1: Bugfix: qconf -aprj was broken. Changed: qconf, sge_qmaster with classic spooling Review: SG JG-2004-02-01-0: Testsuite: Fixed setting of scheduler configuration. Shutdown qmaster separately, after execds and scheduler have been shutdown. Do not try to shutdown sge_commd's in 6.0 systems. CR-2004-01-31-1: Bugfix: removed debug output to stdout Changed: qconf CR-2004-01-31-0: Enhancem.: shadowd now supports new communication library Changed: shadowd Review: pending SG-2004-01-31-1: Bugfix: fixed qconf -de. A exec host can be removed, even so the host can not be resolved. Changed: qconf Issue: 566 Review: CR JG-2004-01-31-1: Testsuite: Fixed a number of installation problems caused by changed objects (queue, schedd_conf) and changed architecture names. JG-2004-01-31-0: Enhancem.: qmake can either use a parallel environment for the execution of make rules, or create an individual qrsh job for each make rule (dynamic allocation). Issue: 338 Changed: qmake, qmake.1 Review: SG MD-2004-01-31-1: Bugfix: - Removed conflict Changed: - sge_config MD-2004-01-31-0: Bugfix: - Change DB install procedure -> only standalone installation, removed from autoinstallation - changed berkeley rc script, supports multible spooling homes now - added shutdown functionality to sgebdb - introduced bdbhomes file - additional module sge_config_berkeley.sh - autoinstall default configfile removed -> the -auto option requires a path to config file - pkgadd rights - typos and cleanups Changed: - sge_config SD-2004-01-30-2: Bugfix: Fixed compile problem Changed: qmon SD-2004-01-30-1: Enhancem.: Removed user_sort from sched_conf(5) Changed: sched_conf(5) Issue: 636 SD-2004-01-30-0: Bugfix: - Removed sort_by_share option from the queue_sort_method - Removed resource_capability_factor from host configuration - Removed oticket and fshare from queue configuration Changed: qmaster, schedd Issue: 643 SG-2004-01-30-0: Bugfix: - fixed qstat -j usage output for pe jobs - fixed qacct -j output. It parsed the acc. file wrong. Changed: qstat, qacct AS-2004-01-30-1: Change: Change architecture names according to Issue #644 Issue: #644 EB-2004-01-30-0: Enhancem.: Changed some man-pages concerning CQs SG-2004-01-30-1: Bugfix: The vmem usage did not decrease anymore. Issue: 635 bugtraq: 5019601 Changed: execd Review: AH SG-2004-01-30-0: Bugfix: - fixed qstat -j usage output for pe jobs - fixed qacct -j output. It parsed the acc. file wrong. Changed: qstat, qacct DT-2004-01-30-2: Bugfix: Minor compile bugs Changed: qsub CR-2004-01-30-0: Bugfix: host aliases file support (part III) - host aliasing was not active when qmaster rereads objects at startup. - reresolve qualified hostname for each qmaster thread at startup. - inst_sge now uses the SGE alias name for creating startup configuration. Changes: qmaster DT-2004-01-30-1: Bugunfix: Rolling back ill conceived change. Changes: qmaster DT-2004-01-30-0: Bugfix: Forgot a header Changed: JAPI MD-2004-01-30-0: Enhancem.: - Added standalone DB installation funktionality Bugfix: - Smoother remove of execd's and queues - deactivated /etc/services entries are not ignored anymore Changed: -sge_config CR-2004-01-29-1: Enhancem.: Added support for /etc/services tcp entries: "sge_qmaster" and "sge_execd" DT-2004-01-29-0: Enhancem.: Added -sync y|n option to qsub to enable waiting for the job end. Extended JAPI to allow: o listening for job start events o starting the event client separately from japi_init() o killing jobs left running in the session from japi_exit() o faster shutdown of the event client o finer grained locking o setting the program name to register with GDI Added DRMAA to prognames. Ported qsub completely over to JAPI. Cleaned up test_drmaa a little. Bugfix: Fixed minor memory management and compilation issues in qmaster and comm lib. Session state was not being included on spooled events. Fixed various minor issues with comments. Fixed qstat compile problem. Changed: DRMAA, JAPI, qsub, qmaster, qtcsh, sge_prog.*, qstat, sge_gqueue.c, comm lib, docs Review: SG, AD pending, AH AD-2004-01-29-0: Bugfix: Qmaster must not invoke 'sge_run_as_user()'. Changed: qmaster JG-2004-01-29-1: Bugfix: When queue instances were spooled as a result of a hostgroup modification, a wrong qinstance name was used for spooling. This resulted in missing queue instances after a qmaster shutdown and restart. Enhancem.: Reduced number of attributes that are spooled for cluster queues and queue instances. Review: SG MD-2004-01-29-1: Bugfix: command (eg. qstat) crashes, if -u -q option get an empty argument Issue: #600 Bugtraq: #5018733 Changed: qstat, qalter, qhost, qmod, qsub Review: EB JG-2004-01-29-0: Bugfix: If a loadsensor did output to stderr, the loadsensor blocked when the pipe buffer between loadsensor and execd was full. Changed: all (libuti) Issue: 632 Review: SG MD-2004-01-29-0: Bugfix: - fixed sgeee option in sge_config, distinst - changed comment in sge_job_master.c Changed: sge_config, distinst SG-2004-01-29-0: Enhancem.: Added qstat -explain E support. Review: JG Changed: qmaster, qstat CR-2004-01-29-0: Enhancem.: - host aliases file support (part II) SD-2004-01-28-1: Enhancem.: - Set default dynamic priority range (ptf_min_priority, ptf_max_priority) for Linux to nice values 0-20 Changed: sge_execd SD-2004-01-28-0: Enhancem.: - Changes to allow user-based functional policy and user-based share tree policy to implement Grid Engine 5.3 user_sort capability - Added support for automatically creating GEEE user objects - Added support for "default" share tree node for users without projects Issue: 636 Changed: sge_qmaster, inst_sge, sge_config, qmon, sge_conf(5), sge_pe(5) AD-2004-01-28-3: Bugfix: 'startup heartbeat' went into wrong directory. Changed: qmaster AD-2004-01-28-2: Bugfix: Did add 'startup heartbeat'. Changed: qmaster JG-2004-01-28-1: Enhancem.: Added queue state and consumable information to the reporting file. Review: SG CR-2004-01-28-1: Enhancm.: - host aliases file support (part I) - TODO: activate host aliases file support EB-2004-01-28-0: Enhancem.: Changed some man-pages concerning CQs MD-2004-01-28-2: Bugfix: qalter lacks option -dl Issue: #599 Bugtraq: #5018726 Changed: sge_qmaster, qstat, qalter Review: EB AD-2004-01-28-1: Bugfix: - all files are now created under the admin user UID/GID - the temporary message file in '/tmp' is no longer used. - qmaster lock file is removed on startup. Changed: qmaster MD-2004-01-28-1: Bugfix: - changed option -s to -sm (install shadow master) Changed: sge_config_common.sh, sge_config MD-2004-01-28-0: Bugfix: - spooling without rpc-server damaged berkeleydb configuration -> no spooling - default domain input didn't work Changed: sge_config_common.sh, sge_config_qmaster.sh, sge.conf, execd_template, master_template rpc_startup_template AD-2004-01-28-0: Bugfix: 'inc_qmaster_heartbeat()' should only be called from within the heartbeat event handler. Changed: qmaster DT-2004-01-28-0: Bugfix: Fixed install script wrt new SGE_EXECD_PORT variable Changed: inst_sge CR-2004-01-28-0: Bugfix: - correct commlib shutdown for qmon - removed unnecessary cl_commlib_shutdown_handle() call in sge_terminate() - japi: when event client thread deregisters from event client interface all open communication handle are closed. - commlib: made some mutexes static - fixed problem when cl_com_setup_commlib() and cl_com_cleanup_commlib() are called more than once. - reworked memory free problem(s) when cl_com_create_handle() fails. Changed: all binaries JG-2004-01-28-0: Bugfix: Fixed the build (some files were dos files). MD-2004-01-27-0: Bugfix: - setting productmode sgeee to default - bugfix logfile creation - path generation Changed: sge_config, sge_config_common.sh SG-2004-01-27-0: Enhancem.: - added a XML print facility for CULL lists. - extended qstat to print XML - comments, qstat man page update, and qstat schema files are still missing. Issue: 634 Changed: qstat Review: JG CR-2004-01-27-2: Enhancem.: - modified default communication timeouts - added comments AD-2004-01-27-0: Bugfix: 'inst_sge' does now check for local Berkeley DB spool directory on Linux. Changed: dist CR-2004-01-27-1: Bugfix.: - Tight integration (qrsh -inherit) (supports now new communication lib) - Execution daemons now have a socket port (SGE_EXECD_PORT) - Integrate SGE_EXECD_PORT in (old) install script and testsuite Review: JG CR-2004-01-27-0: Enhancem.: Produce compile error when there is no define in aimk for sge_gethostbyname() and sge_gethostbyaddr(). Review: AA JG-2004-01-27-1: Bugfix: Fixed aimk LFLAGS for HP11. Do not recommend use of --enable-posixmutexes for the build of Berkeley DB. Do not use f(un)lockfile on DARWIN6 in flatfile spooling. Review: SG JG-2004-01-27-0: Bugfix: Fixed aimk LFLAGS for HP1164. Fixed a number of build problems for tru64. Review: EB SG-2004-01-26-0: Enhancem.: Introcuded a "-t" option in qdel, qmod, qalter to work on array taskes. Changed. qalter, qdel, qmod Review: DT AD-2004-01-26-6: Bugfix: Make sure, 'cl_com_cleanup_commlib()' is invoked only once during sge_qmaster shutdown. Changed: qmaster, gdi Review: CR AD-2004-01-26-5: Bugfix: Did remove redundant 'sge_switch2admin_user()' and 'sge_switch2start_user()' calls. Changed: spool/classic Review: JG AD-2004-01-26-4: Bugfix: Did remove redundant 'sge_switch2admin_user()' and 'sge_switch2start_user()' calls. Changed: gdi Review: AA JG-2004-01-26-1: Enhancem.: Make Berkeley DB spooling the default. Bugfix: Fixed spooling problem with the creation of queue instances. Bugfix: Trash load values after reading exec hosts from the spooling database. Review: AD EB-2004-01-26-0: Enhancem.: Changed some man-pages concerning CQs AD-2004-01-26-3: Bugfix: 'sge_do_log()' does no longer change the effective UID before and after logging. Changed: uti CR-2004-01-26-0: Bugfix: Fixed build for ALINUX Review: AD AD-2004-01-26-2: Bugfix: Did remove redundant 'sge_switch2admin_user()' and 'sge_switch2start_user()' calls. Changed: qmaster AD-2004-01-26-1: Bugfix: Do communication setup after qmaster setup. Changed: qmaster JG-2004-01-26-0: Bugfix: Fixed a problem with berkeley db spooling of queue instances. Use mutexes in qmaster reporting instead of locks. Review: AD AD-2004-01-26-0: Bugfix: Decrease delivery interval for remote events. Changed: qmaster EB-2004-01-25-1: Enhancem.: Added ADOC comment EB-2004-01-25-0: Enhancem.: Changed some man-pages concerning CQs CR-2004-01-25-1: Bugfix: - fixed possible deadlock of communication lib in startup/cleanup phase - cl_com_cleanup_commlib() did not correctly shutdown connected clients CR-2004-01-25-0: Bugfix: - Fixed "failed gdi request" error message when connection timed out. (e.g. editing configuration file for longer than 1 minute) - Event client(s) will reconnect when connection timed out. AD-2004-01-24-0: Bugfix: Did remove unused parameter from 'sge_do_log()' signature. Changed: uti MD-2004-01-24-0: Enhancem.: - Add auto uninstall function for execd - Some documentation Changed: sge_config, sge.conf, sge_config_execd_uninst.sh MD-2004-01-23-0: Enhancem.: - Added new installation script (sge_config) Changed: all installation files, create_settings.sh, distinst, rcp_startup_template Review: pending (AS) EB-2004-01-23-0: Bugfix: qmaster core in cqueue_list_locate_qinstance() AD-2004-01-23-0: Enhancem.: Add error message in case a mutex lock/unlock does fail. Changed: lck AD-2004-01-22-0: Bugfix: 'qconf -km' did not work for the 'LinuxThreads' library (NPTL did work). Changed: qmaster CR-2004-01-22-0: Bugfix: - execd did not signal job - switched of communication log (as default) communication log can be activated with SGE_NGC_DEBUG variable Changed: execd SG-2004-01-22-3: Bugfix: fixed the counting for running jobs in the qevent client. Changed: qevent Review: CR SG-2004-01-22-2: Enhancem.: updated man page for qmod Changed: man1/qmod.1 SG-2004-01-22-1: Bugfix: Removed the renaming of users via qconf. It never worked and should not have been part of qconf. Changed: qconf Issue: 646 bugtraq: 5020131 SG-2004-01-22-0: Enhancem.: 2. part of the qmod adaption Changed: qmod, qmaster EB-2004-01-22-0: Bugfix: CQ related bugfixes - changed defaults auf CQ template - changed verification of pe_list EB-2004-01-21-2: Enhancem.: Cleanup / Comments DT-2004-01-21-0: Bugfix: drmaa_run_bulk_jobs() wasn't submiting the job as an array job. Also cleaned up DT-2004-01-19-0. Changed: drmaa.c, interactive_sched.c Review: AH AD-2004-01-21-1: Bugfix: Handling of 'SGE_DEBUG_TARGET' env var was broken Changed: rmon AD-2004-01-21-0: Enhancem: - 'sge_qmaster' does now handle 'SIGTERM'. - overwork of signal handling. Changed: qmaster AA-2004-01-21-0: Enhancem. qmon cluster queue changes, NOT YET COMPLETED to get the rest of qmon functionality operable again (indexes for lists changed with the intro of cluster queues) Changed: all, sge_attr.c, sge_object.c JG-2004-01-21-0: Bugfix: Categories (job requests) were truncated in the accounting and the reporting file. Fixed build for solaris86. Changed: sge_qmaster Review: SG EB-2004-01-21-1: Bugfix: Handled TODO-Sections concernings CQs Changed: sge_schedd EB-2004-01-21-1: Bugfix: Handled TODO-Sections concernings CQs Changed: sge_schedd EB-2004-01-21-0: Bugfix: Queue selection did not work in combination with cluster queue printing (qstat -g c -q ) Changed: qstat RC-2004-01-20-0: Enhancem.: use sysconf(_SC_NPROCESSORS_ONLN) to get nprocs on Linux. Changed: libs/uti/sge_nprocs.c CR-2004-01-20-1: Enhancem.: communication lib: status information response message reports application status. Changed: ngc lib JM-2004-01-20-3: Bugfix: The aimk configurations for AIX4[123] and AIX51 are changed to use CC=xlc_r and CXX=xlC_r. SHAREDLD (for AIX51) is also set to 'xlc_r'. Changed: aimk Reviewed: RH, AS JM-2004-01-20-2: Bugfix: The setrlimitj() call under SUPER-UX must be surrounded by sge_switch2start_user() and sge_switch2admin_user() to be effective. Changed: daemons/shepherd/setrlimits.c Reviewed: SD CR-2004-01-20-0: Cleanup: - enabled threads for communication in qmaster - hostname resolving (still no alias file support) - communication log can now be switched on/off with environment variable "SGE_NGC_DEBUG" - sge_gethostbyname() and sge_gethostbyaddr() now always return a copy of the hostent struct JM-2004-01-20-1: Enhancem: Revise japi.c to use rand_r() instead of (s)random() which are not reentrant. 'id' is changed from long to int to suit rand_r(). Changed: libs/japi/japi.c Reviewed: AH EB-2004-01-20-1: Enhancem.: new switch "qconf -sick" Changed: qconf AD-2004-01-20-0: Enhancem: Did add 'sge_log_tee' script. Changed: util JM-2004-01-20-0: Adding entries JM-2004-01-19-[0-5] to Changelog SG-2004-01-20-0: Bugfix: fixed qmod and removed lWriteListTo from the code Changed: qmaster, qmod EB-2004-01-20-0: Enhancem.: core implementation for qconf -sick - cleanup - TODO items JM-2004-01-19-5: Bugfix: The trailing ',' is removed from the listed files. Compilers generally warn about it, but the IBM compiler flags it as an error. Changed: libs/sgeobj/sge_schedd_confL.h, daemons/qmaster/sge_reporting_qmaster.h Reviewed: RC (for libs/sgeobj/sge_schedd_confL.h) JM-2004-01-19-4: Bugfix: The 'name' variable is moved outside of the '#ifdef SGE_LOADMEM' block since it is used for SGE_LOADMEM and SGE_LOADCPU. Also, it is explicitly initialized to NULL. Changed: gridengine/source/libs/utilbin/loadcheck.c JM-2004-01-19-3: Enhancem: Support added for getting load average info for AIX51 (missed the .h file). Changed: gridengine/source/libs/uti/sge_getloadavg.h JM-2004-01-19-2: Enhancem: Support added for getting load average info for AIX51. Changed: gridengine/source/libs/uti/sge_getloadavg.c JM-2004-01-19-1: Bug Fix: For SUPER-UX, the total CPUs calculation is no longer based on RSGs (resource group) but on 1) syssx(), or 2) RBs (resource block). The RB method is added in case syssx() is not available (e.g., older SUPER-UX version). Changed: libs/uti/sge_nprocs.c JM-2004-01-19-0: Bug Fix: For SUPER-UX, the total memory calculation is now based on RBs (resource block) instead of RSGs (resource group). Changed: libs/uti/sge_loadmem.c DT-2004-01-19-0: Bugfix: Fixed scheduler crash when an array job with -now y could not be completely scheduled. Now if an array job with -now y cannot be completely scheduled, it is completely removed. Changed: schedd, sched lib Review: SG SG-2004-01-19-1: Bugfix: qhost -F -h Name did not display global consumables Changed: qhost Issue: 607 bugtraq: 5020134 SG-2004-01-19-0: Enhancem.: qmod supports jobnames with new commands. The old ones are still there and work on queues and job ids. The new ones end with j and work on job ids or job names. Or the end with q and work on queue names. Man pages are not changed yet. Changed: qmod, qmaster Review: pending AD-2004-01-19-1: Bugfix: Spooling trigger handler was not registered. Changed: qmaster EB-2004-01-19-2: Bugfix: several bugfixes - fixed "test_drmaa ST_DRMAA_JOB_PS " - communication between master and execd was broken qdel, qmod ... work properly again - removed obsolete files - fixed CQ-TODO-Sections AD-2004-01-19-0: Bugfix: Debug level (dl) scripts now set "SGE_ND" env var, if needed. Changed: util CR-2004-01-19-2: Cleanup: commlib now uses sge_gethostbyaddr() and sge_gethostbyname() from util library CR-2004-01-19-1: Bugfix: fixed memory leak in sge_gethostbyaddr() and sge_gethostbyname() Changed: lib util Review: DT EB-2004-01-19-1: Bugfix: non-existing complex attributes were not rejected during cqueue modification Changes: sge_qmaster, qconf CR-2004-01-19-0: Bugfix: - changed debug level of communication system from TOP_LAYER to COMMD_LAYER - removed not necessary calls to not thread save function inet_ntoa() and changed thread safety comments Changed: qmaster, util lib, gdi lib (all binaries) EB-2004-01-19-0: Bugfix: qstat -g xxx did not show error message for characters which are not allowed Changed: qstat AD-2004-01-18-0: Bugfix: Accounting file has not been written, because reporting trigger event handler was not registered. Changed: qmaster RC-2004-01-17-1: Cleanup: Removed dead code on MacOSX (already #if 0 out) Changed: sge_getloadavg.c RC-2004-01-17-0: Enhancem.: Turned on optimization flag, defined _REENTRANT on SUPER-UX (for John) Changed: aimk Review: RC SG-2004-01-17-3: Bugfix qsub -l BOOL_NAME=VALUE needed the value. Changed: qsub Issue: 618 Review: pending SG-2004-01-17-2: Bugfix: type mismatch in sge_security (pointer instead of unsigned long) DT-2004-01-17-0: Bugfix: Added function to add two lists together while treating them as sets, i.e. element uniqueness is maintained. Also changed exec_job to add the SGE_TASK_* env vars even if they already exist. Issue: 609 Changed: sge_exec.c, sge_var.* Bugfix: Fixed a couple memory issues in execd Changed: dispatcher.c JG-2004-01-17-0: Bugfix: Script submission was broken in qrsh. -C option was ignored for script submission in qrsh. Manpage submit.1: -C option is allowed for qrsh. Issue: 463 Changed: qsh, submit.1 Review: EB SG-2004-01-17-1: Bugfix: scheduler died, when it got events and wanted to print warnings Changed: scheduler SG-2004-01-17-0: Bugfix: Forgot the change the delimiter, when a job name is changed via qalter with a job name. Changed: qmaster, qalter AD-2004-01-17-1: Bugfix: Initialization cleanup Changed: qmaster AD-2004-01-17-0: Bugfix: - 'sge_get_confval_array()' is now MT safe. - 'sge_get_confval()' is now MT safe. - 'sge_readpid()' is now MT safe. Changed: uti SG-2004-01-16-1: Enhancem.: A manager can delete specific jobs from other users or all jobs. Changed: qmaster, qdel, man pages AD-2004-01-16-1: Bugfix: Did fix build on glinux Changed: qmaster AD-2004-01-16-0: Bugfix: Did fix initialization issues Changed: qmaster CR-2004-01-16-1: Bugfix: - JAPI: ensure japi_init()/drmaa_init() does not return before initial event arrived - new formal prog names: "JAPI" and "JAPI_EC" as default program names for uninitialized japi clients ("unknown" endpoint id's). - removed unallowed free() call in test_drmaa binary DT-2004-01-16-0: Bugfix: Fixed uninitialized read in new commlib on Solaris 9 Changed: cl_host_list.c SG-2004-01-16-0: Bugfix: the slots attribute could not be overwritten on host or global level. This has been changed. Changed: scheduler Review: AH CR-2004-01-16-0: Testsuite: prefer solaris64 architecture for building dependencies in compile_source procedure AD-2004-01-15-2: Bugfix: Did fix build on glinux Changed: qmaster, lck AD-2004-01-15-1: Bugfix: - Did apply patch from John Marshall for SUPER-UX Changed: utilbin AD-2004-01-15-0: Enhancem: - MT version of sge_qmaster. - added locks for MT saftey Review: TBD Changed: qmaster, evm, gdi, lck, sgeobj, common SG-2004-01-15-2: Enhancem: - changed delimeter for the reporting file (new ":") - changed the allowed characters in a job name to the common object name specification - qsub -l time requests are transformed into a value based on seconds. This means, that we have no ":"in the accounting and reporting files anymore. Review: JG Changed: qmaster SG-2004-01-15-1: Bugfix: fixed qdel -uall and qdel -u NAME Review: JG Changed: qdel, man pages SG-2004-01-15-0: Bugfix: fixed the build CR-2004-01-15-3: Bugfix: - code clean up (function names) CR-2004-01-15-2: Bugfix: - fixed event client where subscription ( abort() error because a used list was not subscribed) - set synchron receive timeout of communication lib to 2 * event timeout CR-2004-01-15-1: Enhancem.: - new function: cl_com_set_synchron_receive_timeout() - some code clean up (function names) JG-2004-01-15-1: Bugfix: Schedd aborted when scheduling a parallel job. Review: SG Testsuite: Updated access_lists and tight_integration test. CR-2004-01-15-0: Bugfix: removed dummy objects in commlib makefile JG_2004-01-15-0: Enhancem.: Separated flatfile spooling base functions from framework implementation. Use flatfile spooling for qconf -mp/-Mp as preparation for a complete switch to flatfile spooling in qconf. Review: SG DT-2004-01-15-0: Bugfix: Fixed build on HP11 Changed: aimk RC-2004-01-14-0: Bugfix: - Fixed the Super-UX build CR-2004-01-14-3: Bugfix: - removed double include files - setting loglevel of new communication lib (NGC) to CL_LOG_WARNING - modified commlib_v1.0 Makefile for link dependencies - changed some communication log messages CR-2004-01-14-2: Bugfix: shutting down execd can hangup execd in endless loop CR-2004-01-14-1: Bugfix: rcsge startup_template still want to shutdown commd CR-2004-01-14-0: Testsuite: update install_core_system test for new communication lib. SG-2004-01-14-1: Enhancem: enabled qacct to work with cluster queues, queue instances, and queue domains. Changed: qmaster, qacct, man pages Review: pending SG-2004-01-14-0: Bugfix: fixed some error messages in qstat Changed: qstat DT-2004-01-14-0: Bugfix: Fixed issues finding the script file to parse script options when DRMAA_WD is set. Also fixed mistake in sge_copy_hostent. Again. Also found cleaner fix for bug #587. Changed: DRMAA, sge_hostname.c, read_defaults.c JG-2004-01-14-0: Enhancem.: Made the testprogram from JG-2004-01-13-0 available for all spooling methods. Bugfix: Fixed build dependencies. JG-2004-01-13-0: Enhancem: Reconnect to a BerkeleyDB RPC server after a disconnect. Initialize BerkeleyDB connection in newly created threads. Added a test program doing massive parallel spooling from multiple threads. Review: AD CR-2004-01-13-0: Enhancem.: - Added new communication library - Changed system to use new communication lib - Modified install scripts - WARNINGS: o sec_lib not yet supported o not all features are tested o compile only tested for SOLARIS64 and LINUX Changed: all binaries, install script, aimk, ... CR-2004-01-12-0: CR_COMMD_TAG created DT-2004-01-12-1: Bugfix: Added DLLIB for 3rdParty files Changed: aimk DT-2004-01-12-0: Bugfix: Fixed uninitialized error code in sge_gethostbyaddr Changed: sge_hostname.c SG-2004-01-11-0: Bugfix: Fixed the build DT-2004-01-09-2: Enhancem.: More minor updates to test_drmaa. Updated comments in drmaa.c. Updated testsuite to use new tests in test_drmaa. Updated TODO list in drmaa.html. Bugfix: Fixed memory leak and apparent bug in sge_getme(). Changed: testsuite/apis/drmaa, drmma, test_drmaa, sge_prog.c DT-2004-01-09-1: Enhancem.: Finished making sge_gethostbyname() MT-safe. Also made sge_gethostbyaddr() MT-safe. Updated libdrmaa to match latest spec changes. Updated test_drmaa to match libdrmaa. Bugfix: Fixed qmon and commd compile issues on AIX. Changed: DRMAA, sge_hostname.c, aimk, qmon, commd, sge_time.c SG-2004-01-09-1: Bugfix: The scheduler did not renew its event subscriptions after the qmaster went down and up again. Changed: scheduler SG-2004-01-09-0: Enhancem.: qconf -cq is working again with the cqueues Changed: qmaster Review: EB DT-2004-01-09-0: Enhancem.: OpenSSL is now loaded dynamically on Solaris and Linux via dlopen. For more information please see the comments in sec_lib.c starting around line 170. Bugfix: Certificates are now usable when SGE is installed as or running as a user other than root. Changed: sec lib, aimk, inst_sge, miriad Makefiles, qmaster Review: AA JG-2004-01-09-0: Testsuite: Fixed some installation problems. AD-2004-01-08-1: Bugfix: 'prog_state_t' and 'cull_state_t' has only been allocated for the first thread calling one of the prog and cull state get/set functions. Changed: uti, cull Review: CR SG-2004-01-08-2: Bugfix: The scheduler did not get new queue instances, when the system was started without any queue. Changed: scheduler SG-2004-01-08-1: Bugfix: qsub -N error message reports the rules for job names Changed: qsub SG-2004-01-08-0: Enhancem.: second part of SG-2003-01-07-1 Changed: qstat, qselect Review: EB JG-2004-01-08-0: Enhancem.: Use Berkeley DB RPC server in case spooling to a local filesystem is not possible (e.g. if the sge_shadowd shall be used), adapted installation procedure and testsuite. Bugfix: Fix spooling of sharetree. Review: MD, AD AD-2004-01-08-0: Bugfix: 'log_state_t' has only been allocated for the first thread calling one of the log state get/set functions. Changed: uti Review: CR AD-2004-01-07-3: Enhancem: 'SGE_LOCK' and 'SGE_UNLOCK' now use 'DLOCKPRINTF'. Changed: lck AD-2004-01-07-2: Enhancem: Did add test program for recursive locks. Did update existing lock test programs. Changed: lck AD-2004-01-07-1: Enhancem: Did add '-D_GNU_SOURCE' for Linux. Did replace 'CLK_TCK' with 'sysconf(_SC_CLK_TCK)'. Changed: aimk, util Review: AH SG-2003-01-07-1: Enhancem: added qstat -qs and qselect -qs feature. Changed: qstat, man-pages Review: pending DT-2004-01-07-1: Bugfix: Fixed install for -secure Changed: util/arch_variables, inst_sge Review: AA DT-2004-01-07-0: Bugfix: Fixed dependency problem for -secure with Sun make Changed: utilbin/Makefile AD-2004-01-07-0: Enhancem: Did remove preprocessor-conditional for 'GET_SPECIFIC' macros. Changed: common Review: AH SG-2004-01-07-0: Bugfix: Fixed: SG-2003-12-19-0 The scheduler did not get new jobs or cqueues. Changed: qmaster, scheduler AD-2004-01-01-0: Enhancem: Did add build support for DARWIN 7.* Changed: aimk Review: TBD JG-2003-12-29-0: Enhancem.: qrsh_starter will no longer overwrite the DISPLAY variable if it is already set to enable ssh display forwarding if ssh is used as transport mechanism for qrsh. Changed: qrsh_starter Review: TBD AD-2003-12-24-1: Bugfix: Did apply changes of 'AD-2003-12-24-0' to 'sge_lock.c' Changed: uti AD-2003-12-24-0: Bugfix: Did fix 'AD-2003-12-22-1' and 'AD-2003-12-23-0' Changed: cull uti JG-2003-12-23-0: Enhancem.: Added infrastructure for job logging and basic job logging functionality. Added reporting_params in the global config. Cleanup of sge_conf parsing code. Review: CR AD-2003-12-23-1: Enhancem: Did fix the build Changed: utilbin AD-2003-12-23-0: Enhancem: Thread local storage is now initialized on demand. Changed: cull AD-2003-12-22-1: Enhancem: Thread local storage is now initialized on demand. Changed: uti AD-2003-12-22-0: Bugfix: Did add missing header file Changed: sgeobj EB-2003-12-22-1: Bugfix: fixed bug * installation script should use "queue" object specifier for -?attr options of qconf Review: MD EB-2003-12-22-1: Bugfix: fixed bug * memory leak in sge_qmaster during modification of CQs comlex_values Review: CR EB-2003-12-22-0: Bugfix: fixed bug * memory access error in libs/uti/sge_hostname.c Review: CR EB-2003-12-20-0: Bugfix: fixed bug * honour initial state correctly * object_spec for ?attr options for CQs is "queue" * qconf -shgrp_tree * qconf -shgrp_resolved * more detailed error messages for CQ parsing AD-2003-12-19-0: Enhancem.: Added new monitoring class 'LOCK' Changed: util, lck, rmon Review: AH SG-2003-12-19-1: Bugfix: The fix ensures, that a job task is not used with the -hold_jid flag together and that in qalter only job_tasks are acepted, when they are allowed. Review: AS Issue: 551 bugtraq: 4969825 Changed: qmaster, qalter, submit.1 SG-2003-12-19-0: Enhancem.: Enabled the event master to work with the cluster queues. Added a filter for cqueues and iqueues in the scheduler. Changed: scheduler, qmaster Review: JG JG-2003-12-18-1: Enhancem.: Made Berkeley DB spooling thread safe. Cleanup. Review: AD JG-2003-12-18-0: Bugfix: Fixed spooling of cluster queues in Berkeley DB. Fixed some make dependencies. Review: EB AD-2003-12-18-0: Bugfix: Fixed the build JG-2003-12-17-1: Testsuite: Fixed testsuite problem with empty queue list. AD-2003-12-17-0: Enhancem.: Added new timed event module Changed: qmaster Reviewed: AH JG-2003-12-17-0: Bugfix: Fixed make pe template. MD-2003-12-17-0: Enhancem.: Added option (-log) to infotext Changed: infotext.c Reviewed: CR JG-2003-12-16-0: Testsuite: Made testsuite installation work with cluster queues. EB-2003-12-16-0: Bugfix: fixed bugs * fixed invalid event handling Changed: sge_schedd, sge_qmaster SG-2003-12-15-0: Enhancem.: Added the use of job_names or patterns instead of job_ids Changed: qmaster, qsub, qalter, qstat, qacct, qrsh, qlogin qmod, gresub, qhold, man-pages Issue: 326 EB-2003-12-15-1: Bugfix: fixed bugs * "qconf -aq test" without modification caused error message * no output concerning complex entries during master startup * added new cqueue related funtions for qmon JG-2003-12-15-0: Bugfix: Fixed build for solaris86. EB-2003-12-15-0: Bugfix: removed unused messages EB-2003-12-13-0: Enhancem.: Cluster Queue (merged from CQ_BRANCH) Changed: all Review: AH All changes applied to the CQ_BRANCH have been merged to MAINTRUNC. This also means that (nearly) all work packeges for the second milestone of CQs have been realized. Following part are still missing. They will be done till mid/end of january: * "qconf -cq" does not work anymore * "qconf -explain E" does not show result for queues error state * "\" can't be used as continuation character until now * qacct has not been changed accordingly * help output for all commands has not been changed * GUI * Testsuite * man-pages and how-to's are still missing ... JG-2003-12-10-2: Testsuite: Cleanup SG-2003-12-10-3: Enhancem.: Added a cycle check for job dependencies, when an existing job-net is modified. Changed: qmaster SG-2003-12-10-2: Bugfix: qmaster crashed, when -hold_jid specified a none existing job id (SG-2003-12-09-1). Changed: qmaster SG-2003-12-10-1: Bugfix: qsh and qlogin allowed -hold_jid even so it was not part of the -help output. Disabled -hold_jid and -h option for qsh and qlogin. Updated the man pages to reflect the changes. Changed: submit.1, qsh, qlogin bugtraq: 5020141 SG-2003-12-10-0: Enhancem.: changed the man pages to reflect SG-2003-12-09-1 Changed: submit.1 JG-2003-12-10-1: Testsuite: Fixed some problems with JG-2003-12-10-0. JG-2003-12-10-0: Testsuite: Removed SGE version dependencies in the testsuite installation process, prepared cluster queue changes. Cleanup. Review: CR SG-2003-12-09-1: Enhancem.: Reqular expression are allowed for the switch -hold_jid in qsub, qalter, and qmon. Qstat -j displayes an extra row, showing the requested job dependency. The row for current pending job dependencies will only show job numbers. Changed: qmaster, qstat, qsub, qmon Review: AA, JG Issue: 181 Issue: 471 SG-2003-12-09-0: Bugfix: Scheduler crashed. The bug was introduced by: SG-2003-11-20-0 Changed: scheduler JG-2003-12-08-0: Bugfix: Fixed a bug in BerkeleyDB spooling for ja_tasks. Review: EB SG-2003-12-08-0: Bugfix: Storing a job template in qmon set -hold_jid to the wrong value (always 0) Review: AA bugtraq: 5020139 Changed: qmon SG-2003-12-05-0: Bugfix: Finished the implementation of SG-2003-12-03-2. It was only partly finished and not activated till now. Changed: qmaster, scheduler SG-2003-12-04-2: Enhancem.: changed some function paramters to be const. SG-2003-12-04-1: Bugfix: The enhancement AH-2003-11-11-0 redirected all scheduler config changes to the qmaster messages file. This is changed. Qconf will display the error messages again. Changed: scheduler, qmaster SG-2003-12-04-0: Enhancem.: Removed the deadline ticket computation. Changed: scheduler, qmaster, qmon SG-2003-12-03-3: Bugfix: The enhancement AH-2003-11-11-0 kept the old priority when reporting pjob tickets is turned off in a running system. The priority and mtix are set to 0, when the reporting is turned off. Changed: scheduler SG-2003-12-03-2: Enhancem.: The event master JAT-Type filtering was incomplete. The JB-Type JAT-Type sub-lists are now filtered as well. Enabled JAT-Type filtering in the scheduler Changed: qmaster, scheduler SG-2003-12-03-1: Enhancem.: Updated the event client to use the event client filtering mechanisments. Also limited the subscripted events to the ones, which are really needed. Changed: qevent SG-2003-12-03-0: Enhancem.: added documentation for sge_schedd_conf and sge_event_client JG-2003-12-03-0: Bugfix: Fixed build for HP11. SG-2003-12-02-0: Bugfix.: The enhancement AH-2003-11-11-0 disabled the use REStrings as a complex type. Changed: qmaster, scheduler,... DT-2003-11-28-0: Enhancem.: Made sge_gethostbyname() reentrant in four of the five possible ways depending on the architecture. I'll let AD do the fifth way: mutex locking. I changed sge_getme() to use sge_gethostbyname() and sge_gethostbyaddr(). I removed the misplaced daemons/shadowd/gethostbyname.* files and replaced them with a sge_gethostbyname_retry() function in sge_hostname.c. I fixed a minor DRMAA bug when passing in a NULL rusage pointer, and narrowed the scope of the event client using SG's new interface. I also fixed the build on HPUX-11. I still need AD to add mutex locking to sge_gethostbyname() default case and sge_gethostbyaddr(). I also need to apply platform specific gethostbyaddr_r() synamtecs to sge_gethostbyaddr(). Changed: sge_hostname.*, sge_prog.c, gethostbyname.*, aimk, shadowd.c, DRMAA, and lots of Makefiles. AA-2003-12-01-0: Enhancem.: Scheduler configuration adaptions in qmon Removal of obsolete dialog fields in cluster conf JG-2003-11-28-0: Enhancem.: Added sharelog (sharetree usage) to reporting file. Added thread safe sge_strerror to replace strerror. Added documentation and MT safety information. Review: AD AH-2003-11-27-1: Bugfix: Fix schedd crash in case PE/CKTP are used. Disabled shepherd signal handler logging to fix shepherd deadlock. Review: AD AH-2003-11-27-0: Doc: Document qstat -ext cpu/mem units. JG-2003-11-26-0: Bugfix: Fixed reading of lists in berkeleydb spooling. Fixed validate_list function. AD-2003-11-25-0: Enhancem.: Added test program to check whether signal handling is POSIX threads compliant. Changed: utilbin DT-2003-11-25-0: Enhancem.: Fixed broken build (AIX51) and added proper error handling for drmaa_wait when no usage information is available. Changed: japi.c, drmaa.c, drmaa.h, test_drmaa.c JG-2003-11-25-0: Testsuite: Cleanup of testsuite configuration. JG-2003-11-24-1: Bugfix: Fixed memory leak from JG-2003-11-24-0. JG-2003-11-24-0: Bugfix: Fixed writing of the reporting file. DT-2003-11-24-0: Enhancem.: Added rusage to DRMAA and JAPI Changed: japi.c, drmaa.c, sge_give_jobs.c AD-2003-11-24-0: Bugfix: Fixed the build AH-2003-11-21-1: Bugfix: Fix complex(5) urgency verification problem AH-2003-11-21-0: Bugfix: Fix schedd/qstat/... crash Review: EB DT-2003-11-21-1: Enhancem.: Fixed broken build (AIX51) and added test for usage information from drmaa_wait() Changed: test_drmaa.c DT-2003-11-21-0: Bugfix: Fixed scheduler startup issue -- cleaned up event client changes Changed: sge_process_events.c EB-2003-11-21-0: Bugfix: Fixed broken build SuSE 9.0 x86: Linux 2.4.21 i686 GNU C Library 2.3.2 gcc version 3.3.1 SuSE 7.2 Sparc: Linux 2.2.20 sparc GNU C Library 2.2.4 gcc version 2.95.3 JG-2003-11-21-0: Testsuite: Handling of SGE version dependencies. SG-2003-11-20-0: Enhancem.: - Added the posability to specify what and where conditions for the event and the mirror interface. - Added a lSelectElem method, which does the same as lSelect only that it is limited to one elemen - scheduler does not subscribe the global configuration anymore Review: JG Changed: qmaster, scheduler, drama, qevent EB-2003-11-20-0: Bugfix: Broken build AH-2003-11-19-0: Bugfix: The drmaa_run_job()/drmaa_run_bulk_jobs() drmaa_job_template_t * input arguments must be const. DT-2003-11-18-1: Bugfix: Made changes from code review. Mostly stylisitic. Changed: DRMAA & read_defaults.c DT-2003-11-18-0: Bugfix: Fixed broken build Changed: read_defaults.c SG-2003-11-18-0: Bugfix: Fixed the build AD-2003-11-18-0: Bugfix: Fixed the build DT-2003-11-17-1: Enhancem.: Implemented remaining DRMAA attributes. Reworked drmaa_job2sge_job() to reuse the parameter parsing code in parse_job_cull.c, parse_qsub.c, read_deaults.c, etc. Introduced two new options: -cat (job category) and -wd (set working directory). Both options are only used by DRMAA right now, but could easily be added to qsub et al. Also added a test to test_drmaa to test all DRMAA attributes. Changed: libdrmaa.so and common parameter parsing routines Review: AH DT-2003-11-17-0: Bugfix.: -b y in sge_request would override -b n on commandline Changed: qsub AD-2003-11-17-0: Bugfix: Revoke AD-2003-11-14-0. JG-2003-11-17-0: Testsuite: Some minor fixes to make testsuite installation run through without errors. Cleanup: Fixed some build dependencies. Enhancem.: Partial packing of objects for use with the Berkeley DB spooling method. AH-2003-11-14-1: Doc.: Add supplement for AH-2003-11-13-1. AH-2003-11-14-0: Testsuite: Further changes to reflect AH-2003-11-11-0 Testsuite fixes. AD-2003-11-14-0: Bugfix: Value of 'MAXID' was to large to be used in conjunction with 'u_short'. Changed: commd CR-2003-11-13-0: Testsuite: testsuite adaptations for (sgeee) install_core_system AH-2003-11-13-1: Doc.: Update man pages to reflect AH-2003-11-11-0 changes. AD-2003-11-13-0: Bugfix: Log messages of level 'LOG_ERR' and 'LOG_CRIT' have been suppressed erroneously under special conditions. Changed: sge_log EB-2003-11-13-0 Bugfix: qmon cores if it should show qtch jobs which where submitted with -h parameter (qtask-file) Review: AA Issue: Bugtraq: 4949917 Changed: qsh, qrsh, qtcsh AH-2003-11-11-1: Testsuite: Apply changes AH-2003-11-11-0 to testuite 'init_core_system' AH-2003-11-12-0: Testsuite: Increased execd delay after commd startup. CR-2003-11-12-0: Testsuite: testsuite adaptations for install_core_system JG-2003-11-11-0: Bugfix: -M options didn't work for interactive commands (qsh, qrsh, qlogin). The man page was incorrect. Bugtraq: 4952236 Issue: 617 Changed: qsh, submit.1 Bugfix: -notify didn't work for qrsh with command. The man page was incorrect: -notify is not available for qsh, qlogin and qrsh without command Bugtraq: 4952767 Issue: 619 Changed: qsh, submit.1 SG-2003-11-12-0: Enhancem.: - Added reprioritize_interval to the man pages - Added share_deadline_tickets to the spool frame work Changed: scheduler, qmaster, man-pages AH-2003-11-11-0: Enhancem.: Initial revision of of Enterprise Edition policy module enhancments introducing urgency scheme as a resource request dependent policy. - removed weight_tickets_deadline from sched_conf(5) - added weight_deadline to sched_conf(5) - added weight_waiting_time to sched_conf(5) - added weight_priority to sched_conf(5) - added weight_urgency to sched_conf(5) - added urgency colunm to complex(5) - added urgency_slots to sge_pe(5) - added -urg switch to qstat(1) to allow for monitoring urgency policy - enhanced qstat(1) -r output to display contribution in static urgency scheme on a per resource basis - moved job deadline related job columns from qstat(1) -ext to -urg output - sort job list always based on combined (ticket policy + static urgency) and new priority - changed qstat(1) to prints only one line for a parallel job and introduced -g t to retain former multi-line output for parallel jobs. qstat(1) -t option implicitely sets -g t Cleanup: libs/sched/sgeee.c Review: EB AD-2003-11-10-2: Cleanup: RMON macros like 'DPRINTF' or 'DTRACE' must not be used without 'DENTER/DEXIT'. Changed: rmon JG-2003-11-10-0: Bugfix: Fixed the build Enhancem.: Added information (host load and consumables) to reporting file. Which variables are to report is configurable through the exec host configuration. AD-2003-11-10-1: Bugfix: Fixed the build Changed: rmon SG-2003-11-10-0: Bugfix: scheduler died. The resource matching caching did not handle error messages correct. Changed: schedd, qmaster AD-2003-11-10-0: Bugfix: Fixed the build Changed: sec_lib, commlib, cull_list AD-2003-11-07-0: Cleanup: Removed obsolete RMON functionality. Preserved local RMON functionality (RMON light). Changed: rmon Review: AH SG-2003-11-07-2: Bugfix: scheduler crashed, when bool resources were used Changed. scheduler SG-2003-11-07-1: Bugfix: removed to many elements from the queue CULL structure Changed: scheduler SG-2003-11-07-0: Bugfix: three sub lists of the job category where freed twice. Changed: scheduler SG-2003-11-06-3: Enhancem.: Limited the copied queue structure in the scheduler to the needed elements. Removed queue structure elements, which were never used. Changed: scheduler SG-2003-11-06-2: Enhancem.: Added a string type to the centries, which allows regular expression matching. Changed: scheduler, qmaster, qstat, man pages SG-2003-11-06-1: Bugfix: qstat -F did not dump out string and host values Changed: qstat SG-2003-11-06-0: Bugfix: qstat -f -l xxx seg. faulted, when xxx was unknown. Changed: scheduler, qmaster, qstat SG-2003-11-05-0: Enhancem.: Updated the resource_attributes document JG-2003-11-05-0: Enhancem.: Made functions job_get_job_key, job_get_key and job_parse_key MT safe. Reporting: Made reporting functions (accounting and new report file) MT safe Switched to Berkeley DB 4.1.25 AD-2003-11-04-0: Bugfix.: 'sge_qmaster -help' did crash. Changed: sge_qmaster SG-2003-11-04-1: Enhancem.: updated the testsuite to reflect the latest schedd conf changes. Changed: testsuite Review: CR SG-2003-11-04-0: Bugfix: scheduler configuration: - param setting of NULL kills qmaster - usage reporting was switched off - illegal params had been stored Review: AA Changed: all SG-2003-10-31-1: Enhancem.: - Moved the reprioritized flag from the execd_param to the scheduler configuration. - Removed the sgeee_scheduling_interval - Improved the running ticket orders (reduced the size) - ensured, that qmon is still working Changed: qmaster, execd, schedd, qmon SG-2003-10-31-0: Enhancem.: Introduced the "-Msconf fname" switch in qconf to read configuration from a file. Changed: qconf JG-2003-11-03-0: Enhancem.: Check version of BerkeleyDB. Some cleanup and improvement of the dstring implementation, test program for dstrings. Issue: 265 AD-2003-11-03-0: Bugfix.: Using DENTER/DEXIT from within a signal handler does lead to a deadlock. Changed: sig_handlers.c AD-2003-10-30-1: Bugfix.: MT initialization for 'qmon' was missing. Changed: qmon AD-2003-10-30-0: Bugfix.: MT initialization for sge_shepherd was incomplete. Changed: sge_shepherd SG-2003-10-29-0: Bugfix.: - SGE: sched config read in error on qmaster startup - SGE: no jobs were scheduled, fixed - SGE(EE): removed error message: "object_type_free_master_list: has no master list to modify" Changed: schedd, qmaster JG-2003-10-29-0: Bugfix: bootstrap file was owned by root - it has to be owned by the admin_user in admin_user systems. AD-2003-10-28-3: Bugfix.: MT initialization for GDI was incomplete. Changed: gdi AD-2003-10-28-2: Bugfix.: MT initialization for 'commdcntl' was incomplete. Changed: commdcntl AD-2003-10-28-1: Bugfix.: MT initialization of bootstrap code was missing. Changed: japi AD-2003-10-28-0: Bugfix: Fixed the build. Modified shared objects link flags for HP11. Changed: aimk JG-2003-10-27-0: Cleanup: Removed code writing the statistics file. AD-2003-10-27-1: Bugfix: Fixed build for AIX51 AD-2003-10-27-0: Bugfix: Fixed 'commd' Makefile. 'libsec' was required, even if built without '-secure'. AD-2003-10-26-0: Bugfix: fixed the build Added MTFLAGS for HP11 and AIX51 Changed: aimk AD-2003-10-25-0: Bugfix: fixed the build AD-2003-10-24-1: Enhancem.: Added call to 'sge_mt_init()'. Changed: qstat AD-2003-10-24-0: Enhancem.: SGE is now build in multi threaded mode. Changed aimk and Makefiles. Added initialization code for thread local storage. Changed: aimk, Makefiles, library code Review AH EB-2003-10-24-0 Bugfix: broken build SG-2003-10-24-2: Enhancem.: updated the testsuite to deal with the new configuration settings Changed: testsuite Review CR SG-2003-10-24-1: Bugfix: The error message was wrong, when a resource was removed, which was referenced in a pending job. Changed: scheduler, qmaster Review: JG SG-2003-10-24-0: Enhancem.: Second part of the scheduler configuration cleanup. The scheduler configuration is only accessable via access functions. The Master list is only visiable within the sge_schedd_conf module. Qmon, and the function documentation needs to be updated. Changed: all, man-pages Review: JG JG-2003-10-24-0: Bugfix: An invalid time_event was generated, if the spooling method didn't provide a trigger function. JG-2003-10-23-0: Enhancem.: Write accounting file only in fixed intervals, buffer new accounting entries between intervals. Call triggers (spooling and accounting) via the time_event interface. AA-2003-10-21-0: Enhancem.: adaption of complex config dialog - support of column sorting added by clicking the label button Changed: qmon, qconf Review: EB, SG JG-2003-10-20-0: Bugfix: fixed installation started as non root user. SG-2003-10-17-2: Bugfix: fixed the build SG-2003-10-17-1: Enhancem.: First part of the scheduler configuration cleanup. All copies of the scheduler configuration within the scheduler are removed. The scheduler works via access functions directly on the scheduler master configuration. A validate function is added to ensure a correct configuration settings Changed: qmaster, schedd, qmon, qstat,... Review: JG SG-2003-10-17-0: Bugfix: - qsub -w e killed the qmaster - qsub -w e did not accept any job. Changed: qmaster, schedd Review: JG JG-2003-10-17-0: Enhancem.: Berkeley DB spooling: Added trigger function that checkpoints the transaction subsystem in regular intervals and clears the transaction log in regular intervals. Make qmaster call the spooling framework trigger function. EB-2003-10-16-0: Bugfix: qconf -mattr ? produced core AA-2003-10-16-0: Bugfix: In case that a deletion of a resource attribute failed, was the resource deleted anyway. This is fixed now. Qconf -mc reported incomplete messages. Changed: qmon, qconf, qmaster Review: SG JG-2003-10-15-2: Enhancem.: Spooling framework: - added trigger and transaction interface - added validate_list interface and default function - updated all spooling methods to use validate (for single objects) and validate_list (for complete lists after reading lists at startup) JG-2003-10-15-1: Bugfix: Spooling framework: Updated verify function to changed complex entry definition. JG-2003-10-15-0: Enhancem.: Berkeleydb Spooling: - added transaction handling - qmaster: shutdown and restart spooling framework while daemonizing AH-2003-10-15-0: Cleanup: Removed null-statement code in shepherd about truncating of job stdout/stderr. There is no case when we truncate. AH-2003-10-14-0: Bugfix: - matching for -l a= was broken AD-2003-10-14-0: Bugfix: Fixed the build Changed: sge_lock.c AA-2003-10-13-0: Bugfix: - cluster config params went to bootstrap file, removed from dialogue - hostgroup folder added to host dialogue - openssl 0.9.7c - sge_ca fixes from V53 JG-2003-10-13-0: Enhancem.: Bootstrapping: Updated man pages. Bugfix: If directories SGE_ROOT, SGE_ROOT/SGE_CELL and SGE_ROOT/SGE_CELL/common were missing, no error message was output by SGE components. AD-2003-10-13-0: Enhancem.: Added target 'lcklib'. Changed: ./libs/Makefile SG-2003-10-13-1: Enhancem.: Updated thre resource_attributes.txt document. SG-2003-10-13-0: Bugfix: Fixed the build EB-2003-10-11-0: Enhancem.: Started with STAT_SUM CQ-WP - Moved code which changes QI states into new module - Added dynamic QI attributes - Changed spooling code for QI JG-2003-10-10-0: Enhancem.: Moved product mode to bootstrap file. Bugfix: Fixed the build (new solarisx86 compiler is more critical). AD-2003-10-10-0: Bugfix: Fixed the build Changed: msg_sgeobjlib.h AD-2003-10-09-1: Bugfix: Fixed the build Changed: sge_hgroup_qmaster.c AD-2003-10-09-0: Bugfix: Changed shared object compile options for Darwin. Changed: aimk EB-2003-10-09-0: Bugfix: Changes concerning CQs-GDIBASE package - fixed several bugs - modifications due to review with SG - added/changed comments - moved several functions JG-2003-10-09-0: Enhancem.: Bootstrapping: Added qmaster_spool_dir and binary_path to bootstrap file to make scripts and shadowd work for spooling methods other than classic. Updated testsuite to read bootstrap file in case of 6.x systems. SG-2003-10-09-1: Review: Some comments on EBs' first CQ work package SG-2003-10-09-0: Bugfix.: The share tree computed pending job tickets wrong, when the sharetree is not "balanced" (for example: only one branch under the root has pending jobs). Changed: scheduler AD-2003-10-08-1: Bugfix: Provided an initializer for several global variables. Removed duplicate definitions of global variables. AD-2003-10-08-0: Bugfix.: Fixed the build Changed: msg_utilbin.h AD-2003-10-07-0: Bugfix.: 'UTILIB_DRMAA_OBJS' did contain 'config_file.o' two times. This caused link problems later on because of multiple symbol definitions. Changed: lib/uti/Makefile JG-2003-10-07-0: Enhancem.: Bootstrapping: Removed admin_user, ignore_fqdn and default_domain from global configuration. All core components (except shadowd) use the bootstrapping functions. EB-2003-10-07-0: Bugfix.: Fixed bug (SG-2003-09-25-1) EB-2003-10-06-0: Enhancem.: Fixed many bugs concerning CQs-GDIBASE package - spooling - verification code - hostlist modification in HGRP and CQ - HR_Type functions - event handling EB-2003-10-05-0: Enhancem.: create QI due to hgroup changes AD-2003-10-02-1: Bugfix.: Fixed the build Changed: schedd_message.h AD-2003-10-02-0: Enhancem.: Added new library ('liblck') which provides a locking API to be used by lock clients and lock service providers. Review: AH Changed: new SG-2003-10-02-1: Enhancem.: Added resource match caching. The matching results for jobs within one job category is cached. Review: JG Changed: scheduler SG-2003-10-02-0: Enhancem.: Improved the speed of generating ticket orders by 12%. Review: JG Changed: scheduler JG-2003-10-02-0: Enhancem.: Bootstrapping: All configuration data created by inst_sge is written through spooldefaults, a binary based on the spooling framework. The bootstrapping data (spooling information, ignore_fqdn and default_domain) is written to $SGE_ROOT/default/common/bootstrap. Bootstrapping data is still duplicated in the global configuration. Clients still rely on the existence of the global configuration file, works only with classic spooling. EB-2003-09-30-0: Enhancem.: Bugfixes of enhancements: EB-2003-09-28-0, EB-2003-09-24-1, EB-2003-09-24-0 JG-2003-09-30-0: Enhancem. Implementation of the spooling method berkeleydb. JG-2003-09-29-0: Bugfix: statistics file hostload twice instead of load and virtual memory used. Enhancem.: sge_share_mon option -o now appends to an existing output file instead of overwriting it. In continuous running mode, the file is closed between two output operations. Enhancem.: qrsh: if ssh -t was used as transport, the pseudo terminal created by ssh wasn't attached to the executed command. Review: AA, AS Changed: sge_qmaster, sge_share_mon, qrsh_starter EB-2003-09-28-0: Enhancem.: Bugfixes of enhancements: EB-2003-09-28-0, EB-2003-09-24-1, EB-2003-09-24-0 EB-2003-09-28-0: Cleanup: moved/renamed serveral functions (EB-2003-09-24-1, EB-2003-09-24-0) SG-2003-09-26-0: Enhancem.: Updated the resource_attributes.txt file SG-2003-09-25-3: Bugfix: The tickets are calculated for all pending jobs and not only for the ones, which can be dispatched in a scheduler run. Review: AA Changed: scheduler SG-2003-09-25-2: Enhancem.: Qmaster ensures, that the type for build in attributes are not changed. Review: AA Changed: qmaster, qconf, qmon SG-2003-09-25-1: Enhancem.: All attribute changes are validated now and only valid ones are commited. In all other cases error messages are send to the user. Review: AA Changed: qmaster, qconf, qmon SG-2003-09-25-0: Bugfix: A failed namechange of an attribute resulted in the deletion of that attribute. Review: AA Changed: qmaster, qconf, qmon EB-2003-09-24-1: Enhancem.: qconf -?attr for QCs EB-2003-09-24-0: Enhancem.: qconf -sq for QCs SG-2003-09-16-1: Enhancem.: New document describing the functionality of attributes. Changed: doc/devel/rfe/resource_attributes.txt SG-2003-09-15-6: Bugfix: The running ticket update did not work correctly when the sgeee_scheduling interval was set to a very very long time, a sge_scheduling run was done and the sgeee_scheduling interval was set back to a normal amount of time. The scheduler had the very long sgeee_scheduling interval stored. It got changed after its execution. Review: AS Issue: 605 Changed: scheduler SG-2003-09-15-5: Enhancem.: The sending of running job tickets can be disabled with setting "sgeee_schedule_interval" to "0:0:0". The man pages reflect the new configuration posebility. Review: AS Changed: scheduler, man page SG-2003-09-15-4: Enhancem.: Removed CLASSIC_SGEEE_SCHEDULING functionality and its documentation from the man pages Review: AS Changed: scheduler, man page SG-2003-09-15-3: Bugfix: lGetString was used in two places to retrieve a host name This resulted in a termination of the scheduler. Review: AS Changed: scheduler, and other... SG-2003-09-15-2: Bugfix: When a pending job moved to running, it lost its ticket details. It had only the total amount of tickets. This is changed now, the ftix, sharetree tix and others are included now. Review: AS Issue: 606 Changed: scheduler SG-2003-09-15-1: Bugfix: Fixed a bug in the pending ticket calculation. When two projects have the same amount of shares, they were combined in one functional category, which lead to a wrong job order. The fcategories are now defined by the references to the control structures. Review: AS Changed: scheduler EB-2003-09-15-0: Enhancem.: Base implementation for qconf -sq switch for CQ's AA-2003-09-15-0: Bugfix: fixed sge_ca scripts, conf files, added some additional cl switches for extending, printing and verifying certs Changed: sge_ca script SG-2003-09-15-0: Bugfix.: Fixed the build EB-2003-09-11-0: Enhancem.: Event handling for CQ and QI AA-2003-09-11-0: Bugfix: fixed resizing of load dialog queue control Review: SG Changed qmon SG-2003-09-08-0: Enhancem.: - calculation running job tickets allways - fixed functional ptix calculation - calculating final usage in every run (not only sgeee_scheduling_interval) Review: AH Changed: sge_schedd Optim.: - optimized attribute matching in scheduler during job dispatching. - new semantic for attribute matching: - we have a strict hierachy: global -> host ->queue If it does not match on one of the levels, if is over for the job. - within a level: - a load value can be overridden with a user def. value, when the "==" or "!=" operator is used. - in all other cases the minimum/ maximum is used - the values for attributes with ">", ">=", "<=", "<" can be overridden on a lower level, when the new value is more restrictive than the old one. Review: pending Changed. sge_schedd, qmaster, qmon, qstat, qhost,.... EB-2003-09-09-0: Enhancem.: partially implemented qmasters gdi part for queue instance changed: add/remove/modify EB-2003-09-05-0: Bugfix: Broken build EB-2003-09-04-0: Enhancem.: partially implemented qmasters gdi part for queue instance creation according to CQ changes AD-2003-09-03-1: Cleanup: Rearrangement of high level message handling code. Changed: sge_qmaster Review: AH AD-2003-09-03-0: Bugfix: Fixed build on Mac OS X. Removed unused variable. Changed: sge_qmaster, qconf AH-2003-09-03-0: Doc.: Documented that qrsh are not re-runnable. EB-2003-09-03-0: Enhancem.: partially implemented qmasters gdi part for CQ EB-2003-08-29-0: Enhancem.: partially implemented qmasters gdi part for CQ AD-2003-08-27-2: Cleanup: Rearrangement of several qmaster functions in preparation for the multi threaded version of qmaster. This is an interim commit. Changed: sge_qmaster Review: AH Bugfix: Handling of commd message priorities was inconsistent. Changed: sge_qmaster Issue: 561 Review: AH SG-2003-08-27-1: Bugfix: Fixed a bug, which I added during some code cleanup with the last commit. Jobs, which did exed the maxujobs limit were ignored during the ticket calculation and kept their tickets from the last calculation. Changed: sge_schedd SG-2003-08-26-1: Bugfix: The scheduler handled strings in a hard request and hard with a soft request differently, when it was defined on host level and overwritten in a queue. Changed: sge_schedd Issue: 589 Review: EB Enhancem.: Phase 2 of dealing with complexes in the scheduler. In this version the scheduler only generetes attributes in the complex lists needed by the job. This is done by two filtering steps. The first filter contains the names of the complexes requested by the job, the second contains all names avaiable at the current test level( global, host, queue). The test is done with the complexes, which satisfy of first filter and second filter. Changed: sge_schedd, sge_qmaster, qstat, qhost,... Review: EB Enhancem.: The SGEEE scheduler needs quite some time to generate and submit the pending job tickets. Therefor a new "schedd_params" parameter ("REPORT_PJOB_TICKETS") as a boolean switch is introduced. When REPORT_PJOB_TICKETS=0 is set, the scheduler resets all pending job tickets to 0 on the qmaster side and does not report any more pending tickets. The default is REPORT_PJOB_TICKETS=1.? Changed: sge_schedd Review: EB AH-2003-08-26-0: Enhancem.: Made libsec thread safe by using OpenSSL threads(3) interface. Always use and deliver shared DRMAA library and header file. Deliver shared linking OpenSSL libraries where available. Link shared against 0.9.7b OpenSSL libraries. Added aimk -secure-static switch to be used for OS architectures where shared linking OpenSSL libraries are not available. Changed sge_ca and sge_ssl.cnf to support more recent OpenSSL naming convention (AA). Review: AA EB-2003-08-25-0: Enhancem.: Added CQ parameter SG-2003-08-20-0: Bugfix: fixed the build EB-2003-08-19-0: Enhancem.: Added CQ parameter SG-2003-08-19-1: Bugfix: - "Qstat -F complex_name" did not filter for the complex_name and displayed all complexes - working on large job lists was very slow due to the hashing. This change was introduced in JG-2002-12-17-0 as an enhancement. The JG code is commented out and can be added again, when the hashing really improves the speed. - Fixed bug: 586. All slots in a queue are used for job dispatching. Changed: qmaster, scheduler, qstat, qhost, qsub, qacct, ... Review: EB Enhancem. - cleanup in "sge_select_resource" and all functions using it. A variable passed in ("ccl") was never used. - changed the way "xxx2scheduler" functions generate their result list. They build a filter now to generate a list, which only contains the needed elements, instead of removing the unused ones in the end. - devided the complex list for deciding if a job is suitable for a queue into two: a host complex list and a queue complex list. Review: EB Changed: qmaster, scheduler, qmon, ... SG-2003-08-19-0: Bugfix: The complex selection dialog did not show all available complexes. Changed: qmon Review: AA AH-2003-08-18-1: Enhancem. Update on tight integration with Sun HPC ClusterTools 5 AA-2003-08-14-0: Bugfix: Host Configuration -> Execution Host -> Add/Modify Exec Host -> Consumables/Fixed Attributs core dump, when QControl hasn't been opened before Changed: qmon Review: SG AD-2003-08-13-0: Cleanup: Removed '#ifdef PW' code. Changed: ./common/sge.h ./sgeobj/sge_ckpt.c ./qmaster/qmaster.c ./qmaster/setup_qmaster.c ./qmaster/sge_host_qmaster. ./execd/execd.c Review: AA AD-2003-08-08-0: Bugfix: Darwin: use 'pid_t' for typedef of 'JobID_t' Changed: sgedefs.h Review: SG-2003-08-05-0: Bugfix: The profiling worked only in the first scheduling run. An error occured in the second run, which turned the profiling of. Changed: sge_schedd Review: AA SG-2003-08-01-1: Bugfix: All queues for a host in qmon changed into the alarm state, when the host complexes for the hast were changed via qmon (only until the next update). Changed: qmon Review: AA SG-2003-08-01-0: Bugfix: - all queues changed into the alarm state, when the complex list was changed via qconf and the qmon queue window was open. - when the qmom queue window and the complex window overlaped and the complex window is closed with commit, than all queues in the area behind complex window changed into the alarm state. Changed: qmon Review: AA AH-2003-08-01-1: Enhancem. Added integration package for tight integration with Sun HPC ClusterTools 5 Enhanced distinst to install tight integration Review: AA SG-2003-07-31-0: Bugfix: global load value not working as load_threshold Changed: sge_schedd, qstat, qmon, qhost Issue: 589 Review: AH AH-2003-07-31-0 CVS: Created VMAIN_STABLE_3_TAG AA-2003-07-30-0: Bugfix: qmon_cplx: selectAttributes callback was missing, CQ_Type list added libs/sgeobj/Makefile delete generated sge_sub_obj.c SG-2003-07-30-1: Bugfix: QMON displayed all queues in alarm state and displayed an error message, when the load threshold were modified. Review: AA AH-2003-07-30-1: Doc.: Documented env vars affecting DRMAA library calls Improved formatting of DRMAA man pages Removed reference to COMMD_HOST from all man pages except for sge_execd and sgecommdcntl (#451) Issue: 451 Review: AA AH-2003-07-30-0: Enhancem.: Ensure drmaa_strerrno() returns NULL for unknown error codes. EB-2003-07-30-0: Enhancem.: CQ APIBASE WP - initial CQ CULL structure - initial gdi-qmaster code - spooling - event handling Changed: all Enable build rule for DRMAA example. Doc.: Finshed DRMAA ADOC headers. SG-2002-07-30-0: Bugfix fixed the build Changed: sge_schedd SG-2003-07-29-0: Bugfix - Corrected the order of pending jobs in case of functional tickets. - added error handling to lGetRef - Implemented a new algorithm the fticket calculation. Depending on the ratio of fcategories and jobs, it can be a lot faster than the previous allgorithm - added a "\n" to a qmaster error message - fixed some errors in libs/sgeobj/sge_boundaries.h. Some id-ranges were used twice. Issue: 585 Review: AH, EB Changed: sge_schedd, qmaster AA-2003-07-29-0: Bugfix.: - cq preparation: qmon adaption compile problem under SOLARIS/86 fixed AH-2003-07-28-0: Doc.: Initial revision of DRMAA man pages AA-2003-07-28-0: Enhancem.: - cq preparation: qmon adaption CR-2003-07-28-0: Testsuite: - fixed error in path_alias test, using qrsh/qsh -now no option to wait for the job to be scheduled. CR-2003-07-25-1: Testsuite: - fixed timing problem when job is not very long in "t" state in scheduler/job_state_handling test CR-2003-07-25-0: Testsuite: - using -now no option for qrsh in qmod reschedule test EB-2003-07-25-0: Enhancem.: GDIBASE: additional sublist type CR-2003-07-23-0: Testsuite: - new procedure: get_complex_version() - fixed problem when adding queue in fast mode - fixed wait_for_unknown_load() to support "-NA-" string instead of a load value of "99.99" JG-2003-07-23-0: Testsuite: Fixed tight integration test for runlevel 1 (qrsh). BD-2003-07-22-0: Cleanup: Remove definition of MAX_GROUPS. It is unused. Changed: common/basis_types.h Review: AS JG-2003-07-22-0: Bugfix: Finished JG-2003-07-10-0: *_RESERVED_USAGE gives wrong values for PE jobs. Cleanup: Cleanup code related to fix JG-2003-07-09-0. Issue: 580 Changed: sge_qmaster, sge_execd Review: EB CR-2003-07-21-0: Testsuite: modified scheduler and throughput test to support new profiling output from scheduler. EB-2003-07-21-1: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 * After removing slots and np_load_avg attribute it was not possible to add the attributes again Changed: sge_qmaster Review: AA JG-2003-07-21-0: Testsuite: Added test functional/usage checking for issue 580. Cleanup of test functional/tight_integration. Review: CR EB-2003-07-21-0: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 hard resource request caused core of qmaster Changed: all Review: AH EB-2003-07-18-0: Enhancem.: Added support for memory and time values in configuration sublists of CQs Changed: sge_qmaster, qconf Review: AH SG-2003-07-18-1: optim. - Optimized function ticket computation. Decreased the amount of cull accesses when the pending jobs are set active and inactive for calculation its tickets. - Fixed a bug (which I added with the last optimization) for calculating the ftickets for running jobs. Review: AH Changed: sge_schedd SG-2003-07-18-0: Enhancem.: - 2 part: additional profiling information about a scheduler run Some profiling code was added to get a better picture of the bottle necks within a scheduler run. - Changed the profiling output to have a leading "PROF:" string in front of each message. Review: JG Changed: sge_schedd EB-2003-07-17-0: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 * job was rejected because requests a queue while queues are configured as requestable * Moved messages into msg_*.h files * Added ADOC comments Changed: sge_qmaster CR-2003-07-17-0: Testsuite: install_core_system can now handle new complex behaviour. SG-2003-07-17-0: Bugfix: Fixed the build Changed: sge_schedd BD-2003-07-16-0: Bugfix: Change FreeBSD arch strings from freebsd-* to fbsd-*. This fixes arch display for FreeBSD i386, ia64, and ppc. Changed: all SG-2003-07-16-1: Enhancem.: additional profiling information about a scheduler run Some profiling code was added to get a better picture of the bottle necks within a scheduler run. Review: Changed: sge_schedd SG-2003-07-16-0: optim. - Optimized functional ticket computation with share_functional_shares on. - Compute fshare weighting factors only when needed. This can be done because the configuration will not change during one sheduling run. - Unset run status only for the queued jobs, which were set running during ftix compution. Review: AH Changed: sge_schedd CR-2003-07-16-0: Testsuite: bugs/issuezilla/193: cleanup function should not run for sge systems < 6.0 EB-2003-07-16-0: Bugfix: Changed code concerning hostgroup modifications via qconf -?attr options Changed: sge_qmaster, qconf JG-2003-07-16-0: Testsuite: Added a testsuite function "dump_ts_config" to dump the testsuite configuration (attribute name, value and description) to a file. EB-2003-07-15-1: Enhancem.: Bugfix for Enhancem. EB-2003-05-15-0 * wrong attribute reference in answer list message of qconf -mattr request Changed: sge_qmaster SG-2003-07-15-0: Bugfix: Fixed the build Some message IDs were used twice Changed: sge_qmaster, sge_scheduler EB-2003-07-14-2: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 * qmaster core: qmon send wrong target info Changed: qmon, sge_qmaster EB-2003-07-14-1: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 * handle suspend_thrsholds correctly Changed: sge_schedd, qstat, qhost AA-2003-07-14-0: Enhancem.: qmon adaption for EB-2003-07-09-0 * dialogue changes that reference complexes Changed: qmon EB-2003-07-14-0: Enhancem.: Bugfix for Enhancem. EB-2003-07-09-0 * handle load_formula in schedd correctly (missing attributes will result in high load) Changed: sge_schedd EB-2003-07-11-0: Enhancem.: Bugfixes for Enhancem. EB-2003-07-09-0 * handle attributes only as forced attributes if forced is specified in centry definition AND if a value is given in complex_values of host or queue * missing complex attribute value (no load value, no host/queue-complex_value) will result in load alarm Changed: sge_schedd, qstat JG-2003-07-11-0: Testsuite: Enhanced tight_integration test to - check setting of NSLOTS and NHOSTS in tasks environment (issue 578) - check shutdown of job when the master task fails (issue 579) - check shutdown of job when a slave task fails - check if job is shutdown when a slave task exceeds a wallclock limit (issue 575) Review: CR JG-2003-07-10-0: Bugfix: *_RESERVED_USAGE gives wrong values for PE jobs. NOTE: vmem for loosly integrated pe jobs is still incorrect. Can't fix it now due to conflicts with complexes change (EB-2003-07-09-0). Issue: 580 Review: EB Changed: sge_execd AH-2003-07-10-0: Enhancem.: In case of event client thread setup problems drmaa_init() returns error reason known by the implementation thread. Implemented drmaa_control(), drmaa_version(), drmaa_get_attribute_names() and drmaa_get_vector_attribute_names(). Implemented timeouts in drmaa_wait() and drmaa_synchronize(). Introduced C module for multi-threading utilities. Always use sge_mutex_lock()/sge_mutex_unlock() to support DPRINTF() output in case of deadlocks - except in rmon module itself. Added mutex 'message_id_view_mutex' in sge_language.c to fix deadlock in case of error during sge_language.c initialization. Testsuite: Added various testsuite tests for the enhancm. Added test for schedd crash bug #545 EB-2003-07-09-0: Enhancem.: Made changes with complexes according to CQ spec. * data structures - removed CX_Type element - now it is necessary to provide lists of type CE_Type to all functions which accepted CX_Type elements previously - removed "value" column from CE_Type - removed "complex_list" from queue/host specification * CE_Type: GDI/Event/Spooling - all requests address one complex attribute * qconf - changed various options (Ac, ac, dc, scl, Mc, mc, sc) * qstat - changed output of -F parameter - "---" if there is no load available * Changed installation procedure * man pages - changed man pages according to CQ spec Following things still have to me done: * qmon modification * testsuite changes Review: AH Changed: all JG-2003-07-09-2: Bugfix: failure of master task of a tightly integrated parallel job did not delete the job. Bugtraq: 4885930 Issue: 579 Review: EB Changed: sge_qmaster JG-2003-07-09-1: Bugfix: NSLOTS and NHOSTS incorrectly set in environment of tightly integrated tasks Bugtraq: 4885906 Issue: 578 Review: EB Changed: sge_execd JG-2003-07-09-0: Bugfix: Slave tasks of tightly integrated parallel jobs exceeding wall clock time were not killed. Bugtraq: 4881949 Issue: 575 Review: EB Changed: sge_execd SG-2003-07-03-0: Bugfix: max_u_jobs could count too many jobs The sge_add_job had multiple exits after the max_u_job counter was increased. This was fixed. For some reason, this feature was disabled in the maintrunk. I enabled it. Bugtrag: 4886026 Issue: 576 Review: AS Changed: qmaster EB-2003-07-01-0: Bugfix: SGEEE: qmaster crashed on qdel of tightly integrated parallel job Bugtraq: 4883714 Issue: 572 Review: JG, (AS) Changed: sge_qmaster CR-2003-06-30-1: Testsuite: qevent -ts option was not working for sge system Changed: qevent AH-2003-06-30-0: Bugfix: Fixed bug "sge_schedd dies...". Issue: 545 Bugtraq: 4859658 Review: EB CR-2003-06-30-0: Testsuite: fixed aftereffect in test for IZ 509. Enhancments in submit_job() procedure caused unexpected error when testing -t option with invalid range specification. AD-2003-06-25-1: Bugfix: spool_flatfile_open_file(): Changed size of buffer for tmp filename from L_tmpnam to SGE_PATH_MAX. Changed: sge_spooling_flatfile.c AD-2003-06-25-0: Bugfix: New implementaion of 'sge_tmpnam()' 'sge_tmpnam' internally used tmpnam. On glibc based systems this provoked a warning. Issue: 289 Review: EB Changed: sge_tmpnam.[ch] (new), sge_io.[ch] CR-2003-06-25-0: Bugfix: disabled infotext error output if SGE_ROOT is not set. Review: AS Bugtraq: 4885719 Issue: 577 Changed: utilib (all binaries) CR-2003-06-18-1: Testsuite: modified drmaa test - disabled test for GE versions < 6.0 CR-2003-06-18-0: Testsuite: added new testsuite version names: - SGE_5.3prep4 - SGEEE_5.3prep4 - SGE_5.3p4 - SGEEE_5.3p4 EB-2003-06-17-0: Bugfix: Checkpointed job was migrated on bad exit status Restarted kernel/app level checkpointing jobs where rescheduled forever Now jobs also exiting with exit status > 128 won't be rescheduled Issue: 174 Review: AS Changed: sge_shepherd SG-2003-06-16-2: Testsuite: Added a test for issue #544 under bugs/issuezilla SG-2003-06-16-1: Testsuite: Added a test for issue #540 under bugs/issuezilla CR-2003-06-16-0: Testsuite: Added l10n support for add_operator() and delete_opterator() AD-2003-06-12-0: Testsuite: Added test for issue #193 under bugs/issuezilla Code has been fixed previsouly IZ: #193 Changed: sge_procedures.tcl Review: JG JG-2003-06-12-0: Bugfix: Binary submission (qrsh default mode) was broken. Review: AA Changed: qsub, qrsh, qsh, qlogin JG-2003-06-10-0: Bugfix: An incomplete resource request (-l =1) caused qmaster to crash. Bugtraq: 4876169 Review: AS Changed: sge_qmaster, qsub, qsh, qhost, qalter, qstat SG-2003-06-10-0: Bugfix: I added a new bug while fixing issue 540 The qmaster creashed in a special error case. Review: AA Changed: qmaster CR-2003-06-05-0: Testsuite: fixed bug in throughput test for old scheduler time info messages without system and user time JG-2003-06-05-0: Enhancem.: The function returning the CULL object type of sublists now is generated from the CULL SGE_LIST definitions. Removed some unused I18N messages. Review: EB JG-2003-06-04-1: Enhancem.: Primary keys for CULL object types are determined from CULL definition, instead of hard coding (function object_get_primary_key). Review: EB SG-2003-05-26-0 Bugfix the queue name was truncated to 10 chars - added a queue name line to the qstat -r option, which displays the full name IZ: #555 Changed: qstat Review: AH SG-2003-05-26-0: Bugfix qstat crashed, when used with the options -r -s z - added the required information to the list IZ: #545 Changed: qstat Review: AH SG-2003-05-22-0: Bugfix: enabled output file names to have a ":" inside for qsub - changed the hostname file definition for the c parameters -e, -o -S - updated the man pages - implemented a hostname resolving, returning an error if the name could not be found. - implemented a check, if the files are only specified once per machine. IZ: #540 Changed: qsub, qsh, qalter, sge_qmaster, qmon Review: AH, JG-2003-06-04-0: Enhancem.: PostgreSQL spooling: - compile time option to switch between synchronous and asynchronous communication - compile time option to suppress transaction handling - table names will be derived from list field names, special handling for some too long field names Cleanup: - removed incomplete implementation of resource requests with ranges Review: EB, AA CR-2003-06-03-0: Testsuite: - delete all jobs before reinitialize system after test failure - bugs/issuezilla/440: fixed parsing problem on hosts with login messages AH-2003-06-03-0: Testsuite: Added DRMAA API tests. Review: JG AH-2003-06-02-0: Enhacem.: Use session filtering with all non-total-update events. Since sgeE_JOB_USAGE and sgeE_JOB_FINAL_USAGE are non-total-update events session event filtering is applied to these events. Don't use job error state for DRMAA jobs (JOB_TYPE_NO_ERROR). Started implementing drmaa_wif*() functions. Made DRMAA work also with GEEE. To facilitate this now only sgeE_JOB_FINISH events (JR_Type) are used to track job exists events in JAPI. Code deriving SGE job templates from a DRMAA job template is now kept in drmaa.c Added automated session key generation to JAPI. Added session restart capability to JAPI. Interface JAPI session restart capability in DRMAA by means of SGE_SESSION_KEY and SGE_KEEP_SESSION env var. Use env var SGE_DELAY_AFTER_SUBMIT to interface with hook for delayed submission. Added new test cases to ensure jobs finish with DRMAA_PS_FAILED state and drmaa_wifaborted()==true if wrong input/output/error file was specified. Review: EB CR-2003-05-28-0: Testsuite: fixed errors when deleting SGE_ROOT directory in compile_source() procedure. JG-2003-05-28-0: Enhancem.: Spooling to PostgreSQL Database for all objects. Database structure is generated from CULL object definitions. Review: TBD JG-2003-05-27-0: Bugfix: Type mismatch in (compiler warning) in featureset. Review: EB SG-2003-05-27-0: Testsuite: - removed error message with maintrunk removed unused macro call "REFERENCED_IN_QUEUE_LIST_OF_CHECKPOINT" Review: CR CR-2003-05-22-0: Testsuite: - enhancments to make testsuite compatible with cvs commit from EB-2003-05-15-0 (PE/CKPT references moved into queue configuration) EB-2003-05-22-0: Bugfix: Changed qtype output in qstat according to man page IZ: #14 Changed: qstat Review: AH EB-2003-05-20-0: Enhancem.: Got rid of 'history' Removed history as described in section qacct(1) and sge_qmaster(8) in specification for Cluster Queues (#14) IZ: #14 Changed: all Review: AH JG-2003-05-20-0 Enhancem.: Database spooling (PostgreSQL): - spooling framework implementation (not for all objects yet) - script to setup database - script to switch between spooling with and without history Event Mirror: When receiving the SHUTDOWN event, do not exit, but set the global variable shut_me_down. Exit will be done by the application. Review: EB CR-2003-05-20-0: Testsuite: - fixed error when installing without root permission in test init_core_system() - reading macro messages file at testsuite startup CR-2003-05-19-0: Testsuite: - enhancments to make testsuite compatible with cvs commit from EB-2003-05-15-0 (PE/CKPT references moved into queue configuration) CR-2003-05-15-0: Testsuite: - new method to cleanup remote spooldirs in install_core_system check - new function startup_execd_raw() to start an execd without startup script AD-2003-05-15-0: Bugfix: aimk (Arch LINUX5/6): added '-lXmu' to XLIBS. Otherwise the '_XEditRes*' functions could not be resolved or linked. Changed: aimk JG-2003-05-15-0: Bugfix: - Installation of SGE(EE) as non root user failed, as an admin user called "none" was created instead of the install user's name. - removed confusing error messages about changing into the spool directory and scheduler configuration not existing IZ: 535 Changed: spooldefaults, sge_qmaster Review: CR EB-2003-05-15-0: Enhancem.: Move queue_list from PE/CKPT to queue configuartion (part 2/2: move pe/ckpt references into QU_Type) Apply changes 8c) to queue/pe/ckpt data structure Change the queue_conf(5)/sge_pe(5)/checkpoint(5) file formats accordingly these changes. Ensure qtype behaviour is changed as described in queue_conf(5) Find more details in the document summarizing the specification for Cluster Queues (#14) IZ: 14 Changed: all Review: JG (AH) AA-2003-05-14-0: Enhancem.: cluster queue prelim qmon changes AH-2003-05-13-0: Bugfix: fix qacct crash in case job maxvmem equals to 0 AA-2003-05-13-0: Bugfix: allow arguments for qlogin etc. in qmon Changed: qmon AD-2003-05-12-0: Bugfix: aimk (platform LINUX6): HAS_IN_PORT_T is defined depending on the libc version. The corresponding test know is able to deal with 2.2.x and 2.3.x libs. Changed: aimk CR-2003-05-09-1 Testsuite: throughput test: use scheduler system + user time instead of wallclock time for measurement. CR-2003-05-09-0 Testsuite: finalized scheduler performance test JG-2003-05-09-0 Bugfix: Broken build AH-2003-05-08-0 Enhancem.: Use commd commproc timeout as maximum for event delivery time. Introduced EV_THROTTLE_FLUSH "busy handling" to facilitate large number of JAPI event clients. Introduced session key based event filtering to facilitate large numbers of JAPI event clients. Added ec_get_id(). Ensure qmaster adds no new event client in case the commproc registering the event client is already associated with an event client. Use latest DRMAA C binding. Introduced sge_getpwnam_r() to fix race condition in drmaa_run_job(). Block SIGPIPE in JAPI library. Introduced intermediate SGE_JAPI_SESSION env var interface in JAPI library to set JAPI session. Introduced SGE_JAPI_EDTIME env var interface in JAPI library to set event delivery time. Introduced SGE_JAPI_FLUSH_DELAY_RATE env var interface in JAPI library to set event throttle flush rate. Use gdi_kill(EVENTCLIENT_KILL) to quicken event client thread disengagemnt on drama_exit(). Do a leave_commd() in thread that calls drama_exit() and in event client thread. Implemented drmaa_synchronize(). Added four drmaa_synchronize() test cases. Added a test case to verify DRMAA reenrolls transparently at commd after a commproc timeout in commd. Review: EB CR-2003-05-08-0 Testsuite: - optimized start_remote_tcl_prog() and execute_func parameter JG-2003-05-08-0 Bugfix: qmake dumped core when starting recursive make calls IZ: #547 Bugtraq: #4860391 Review: AS Changed: qmake AH-2003-05-08-0 Bugfix: Fix reopend bug "qconf gets commd timeout". Improved commd commproc timeout logging. Do only one retry in commlib to transparently reenroll to commd. Issue: 308 Bugtraq: 4753669 Review: AS EB-2003-05-08-0 Bugfix: settings.[c]sh did not set LD_LIBRARY_PATH correctly if multiple GE installations are present Review: AS Changed: dist/util/create_settings.sh CR-2003-05-07-0 Testsuite: fixed error for install_core_system check: don't add complex list "global" if it already exists. EB-2003-05-07-0 Cleanup: wrapper functions to access QU_qtype attribute (part 1/2: move pe/ckpt references intgo QU_Type object) JG-2003-05-06-1 Bugfix: broken build JG-2003-05-06-0 Bugfix: array tasks were not deleted (bug introduced in change JG-2003-04-29-0), some cleanup. Review: EB EB-2003-05-05-0 Bugfix: qhost -F broken Issue: 530 EB-2003-05-02-0 Bugfix: broken build CR-2003-04-30-0 Testsuite: modified scheduler test to create more queue instances JG-2003-04-29-0 Enhancem.: Reset information which fields have changed in qmaster's objects after spooling and sending events. Bugfix: Removed some classic spooling code from qmaster (replaced by calls to spooling framework). Review: EB EB-2003-04-29-0 Bugfix: SGEEE: scheduler dispatches wrong jobs if there are jobs in the list of pending jobs which had an error state or which have been rescheduled previously Issue: 532 Bugtraq: 4847814 Changed: sge_schedd, qstat, qmon Review: AH, (SD) CR-2003-04-28-0 Testsuite: New scheduler test. - testing consumables performance for scheduler CR-2003-04-25-0 Testsuite: Bugfix for spool_array_to_file() when spooling without comment. JG-2003-04-22-0 Bugfix: qrsh (rshd) does not set limits after setting the osjobid. This disabled limits that depend on being in a OS managed job (e.g. on SUPER-UX/SX). IZ: 513 Review: EB Changed: rshd CR-2003-04-16-0 Testsuite: sharetree test: Don't trigger scheduler run when a job is already running in a queue. JG-2003-04-15-0 Enhancem.: Added infrastructure for spooling to PostgreSQL database. Use relative include paths for spooling related includes. Fixed dependencies on spooling libraries. CR-2003-04-14-0 Testsuite: communication test: added version information to test report CR-2003-04-11-0 Testsuite: communication test: added function communication_compare_data_files() to compare test data files. JG-2003-04-11-1 Enhancem.: Pass answer list to functions defined in the spooling framework. Validate / output contents of this answer list after every call to a spooling function. Removed unused I18N messages in various modules. JG-2003-04-11-0 Testsuite: procedure unsuspend_job failed with tcl error Review: CR JG-2003-04-10-0 Testsuite: avoid testsuite timeout during install_qmaster when SGE is installed on a slow NFS mounted filesystem. Review: CR EB-2003-04-10-1 Testsuite: test #524: qdel -u does not delete all jobs of the user Issue: 528 EB-2003-04-10-0 Testsuite: test #526: cannot qalter/qhold/qrls several tasks of same job Issue: 527 CR-2003-04-09-2 Testsuite: communication test: using testsuite_snd/_rcv -tcp-close option for test CR-2003-04-09-1 Testsuite: using of open_spawn_process() may cause problems on NFS mounted testsuite installations. Using open_remote_spawn_process() even for local spawns. EB-2003-04-09-0 Bugfix: cannot qalter/qhold/qrls several tasks of same job Issue: 526 Bugtraq: 4845505 Changed: qalter Review: JG JG-2003-04-09-1 Bugfix: utilbin/spooldefaults didn't write managers and operators in classic spooling. JG-2003-04-09-0 Enhancem.: Enable use of flatfile spooling for qmaster. Changed the installation process to call a special spooling client to create default configurations using the selected spooling method. CR-2003-04-09-0 Testsuite: communication test: - fixed compiler warnings Changed: testsuite_rcv, testsuite_snd BD-2003-04-09-0 Bugfix: Remove accidently comitted qmake binaries. CR-2003-04-08-7 Bugfix: with CR-2003-04-08-4 was also changing SOLARIS and SOLARIS64 - now it only switches to -xO3 for SOLARIS86 Bugtraq: 4847802 Issue: 531 CR-2003-04-08-6 Bugfix: synchronized qmake Makefiles for all architectures - disabled automatic Makefile generation through configure CR-2003-04-08-5 Testsuite: communication test: - new options for test binaries Changed: testsuite_rcv, testsuite_snd CR-2003-04-08-4 Bugfix: using -xO3 instead of -xO4 for SOLARIS86 Changed: aimk EB-2003-04-08-1 Bugfix: qdel -u does not delete all jobs of the user Issue: 524 Bugtraq: 4842878 Changed: qdel Review: JG EB-2003-04-08-0 Bugfix: sge_shepherd does not exit on SIGTERM Issue: 522 Bugtraq: 4844838 Changed: sge_shepherd Review: AH CR-2003-04-08-3 Bugfix: used wrong variable type (int) for return value of lGetUlong() Changed: qmaster.c CR-2003-04-08-2 Testsuite: throughput test: - increased load_thresholds - show filename of test results data in html report CR-2003-04-08-1 Enhancem: new options for test binaries. Changed: testsuite_snd, testsuite_rcv CR-2003-04-08-0 Enhancem: clean rule for drmaa objs Changed: libs/japi/Makefile JG-2003-04-07-0 Enhancem.: - Improved flatfile spooling - removed dependencies of qmaster on classic spooling AH-2003-04-07-0 Enhancem: - implemented drmaa_job_ps() - fix problems with drmaa_wait() operating on individual jobids rather than on DRMAA_JOB_IDS_SESSION_ANY - event client thread reconnects to qmaster during session when qmaster is down - ensure event client does not exit(1) DRMAA library - investigate sgeE_JOB_LIST event to update library session data - ensure low level GDI errors/warnings are not printed but reported in DRMAA diagnosis buffer Review: JG EB-2003-04-06-0 Testsuite: Added new testsuite test for issue 509 Issue: 520 Review: CR EB-2003-04-04-0 Bugfix: jobs suspended due to calendar were handled as running jobs within scheduler (see EB-2003-03-27-0) Review: AH Changed: sge_schedd CR-2003-04-03-0 Bugfix: commd ignored first message for commlib client in case of priority tags messages (used for qmaster <-> execd communication). The first message was ignored when more than one message with the same priority tag was ready to send. Review: AH Changed: commd EB-2003-04-03-1 Testsuite: Added new testsuite test for issue 511 Bugtraq: 518 Review: CR RC-2003-04-03-0 Enhancem: Start checking at argv[1] instead of argv[0] for sge parameters. Bugtraq: 507 Changed: qmake EB-2003-04-03-0 Bugfix: Unable to delete task array job with negative increment Issue: 509 Bugtraq: 4841414 Changed: qsub, qmon Review: JG AH-2003-03-31-0 Enhancem: - introduced iterators in DRMAA C binding to handle vector output arguments - added support for DRMAA bulk/array jobs - imlemented DRMAA_WD/DRMAA_OUTPUT_PATH/ DRMAA_ERROR_PATH DRMAA_INPUT_PATH/DRMAA_JOIN_FILES/DRMAA_JS_STATE - allow parameterizing of JB_cwd to support DRMAA path placeholders $drmaa_incr_ph$ and $drmaa_hd_ph$ EB-2003-03-28-0 Bugfix: sharetree can't be modified Issue: 515 Bugtraq: 4838636 Changed: sge_qmaster Review: AS EB-2003-03-27-1 Bugfix: maxujobs scheduler config functionality is broken Issue: 514 Bugtraq: 4838549 Changed: sge_schedd Review: AH EB-2003-03-27-0 Bugfix: maxujobs does not count suspended jobs or running/suspended jobs which have an hold applied, a new starttime attached or new job dependencies defined after the job has been started. Issue: 511 Bugtraq: 4838595 Changed: sge_schedd Review: AH CR-2003-03-27-0 Enhancem: implemented FIFO for commd. Changed: commd Review: AH AH-2003-03-27-1 Bugfix: qrong qstat -j diagnosis message wrongly indicates not enough PE slots Issue: 415 Bugtraq: 4775325 Review: AS JG-2003-03-27-0 Bugfix: qstat -alarm was broken. Issue: 469, 489 Bugtraq: 481579 Changed: qstat Review: AH JG-2003-03-26-0 Bugfix: spooling integrity improvement was missing for sharetree, projects and users Issue: 501 Bugtraq: 4822746 Changed: sge_qmaster Review: EB AH-2003-03-26-2 Bugfix: NOTIFY_SUSP signal only sent for first suspension of job Isssue: 510 Bugtraq: 4835832 Review: EB AH-2003-03-25-1 Enhancem: Fix broken build for aimk -only-mt -secure JAPI can now be compiled with -secure EB-2003-03-25-0 Bugfix: Multiple job script writes cause ETXTBSY (Text File Busy) Errors Issue: 504 Bugtraq: 4838650 Changed: sge_execd Review: AH AH-2003-03-25-1 Bugfix: GEEE: allow qalter -p for running jobs Issue: 502 Bugtraq: 4824104 JG-2003-03-24-0 Enhancem.: Added aimk switch -purify to instrument code for memory debugging with purify. AA-2002-03-21-0 Bugfix.: corrections for DARWIN port, see 5.3 branch EB-2002-03-20-0 Enhancem.: Review of EB-2002-03-11-0 and EB-2002-03-14-0 - minor modifications Review: AH (JG partially) AH-2003-03-20-0 Enhancem.: - Integrated DRMAA/JAPI build into regular aimk build procedure - Cleanly separated DRMAA and JAPI implementation - DRMAA library now covers all Grid Engine objects files - Renamed test_japi to test_drmaa - Integrated rmon into JAPI to allow it being used in DRMAA clients without interfacing rmon Review: AS AA-2003-03-19-0 Enhancem.: - Darwin port JG-2003-03-19-0 Enhancem.: - cleanup and profiling for libevc - basic profiling for commlib, gdilib - profiling: allow nested calls to the profiling measurement start function within one level Review: EB CR-2003-03-19-2 Testsuite: fixed compiler warnings for testsuite_rcv/snd CR-2003-03-19-1 Testsuite: new test for communication (commd) performance CR-2003-03-19-0 Testsuite: Enhanced create_html_table() to support not set table entries JG-2003-03-18-2 Enhancem.: Improved multilevel profiling in scheduler. Review: EB JG-2003-03-18-1 Bugfix: fixed memory access error in sge logging. Review: CR JG-2003-03-18-0 Bugfix: sge_schedd dumped core on exit. EB-2003-03-17-0 Bugfix: qsh/qrsh/qlogin might core with segementation violation Issue: 508 Bugtraq: 4833346 Changed: qsh Review: JG EB-2002-03-14-0 Enhancem.: Added code for CQ attribute lists - added new objects (ABOOL_Type, AULNG_Type) which will hold host and hostgroup concerned attribute values in an CQ. (Other list will follow soon) - Summarized parsing code - Changed several ADOC comments EB-2002-03-13-0 Bugfix: Core of master if usermapping is diabled CR-2002-03-12-1 Cleanup: new version of gmake Makefile CR-2002-03-12-0 Bugfix: sge_depend was broken for solaris architecures EB-2002-03-11-0 Enhancem.: Implemented hostgroups according to CQ spec. - added new object (HR_Type) which will be used to store host and hostgroup references. - Reimplemented existing hostgroup code. - Changed fileformat of hostgroups - Enabled hostgroup GDI interface - Implemented all hostgroup related code for all qconf switches including -?attr options - Added code for the event handling of hostgroups - Changed code concerning usermapping - Changed fileformat of usermapping - Added new object (ASTR_Type) which will be used to store default/hostgroup/host attributes for cluster queue and usermapping objects - Changed hostgroup(5), usermapping(5) man page - Added ADOC comments - ... CR-2003-03-10-0 Bugfix: distinst failure for -allall option if user has no access to PDF manual file. CR-2003-03-06-0 Cleanup: Removed unused messages from message files (msg_*.h) CR-2003-03-05-1 Bugfix: Missing "return" in main() function reports compile error for solaris86 CR-2003-03-05-0 Testsuite: - only test job environment variable TZ if it is set on qmaster host environment - fixed bug in handle_vi_edit() when 4th parameter string is empty CR-2003-03-04-1 Bugfix: Fixed compiler warnings for glinux systems in qevent.c module CR-2003-03-04-0 Bugfix: Removed not unique message id 49094 CR-2003-03-03-2 Bugfix: - qevent -help worked only for running GE system CR-2003-03-03-1 Enhancem. - new function in sge_dir.h: sge_is_executable() - qevent client: client can now trigger script execution on job end events. - testsuite extension for new qevent -ts option CR-2002-03-03-0 Bugfix: Removed unused variables in scheduler.c AH-2002-02-28-1 Enhancem.: Ensure drmaa_exit() can be called while a second thread does a drmaa_run_job() or drmaa_wait(). Enhanced test_japi into a test application providing means to do multiple tests. Implemented drmaa_strerror(). Review: CR AH-2002-02-28-0 Enhancem.: Event client was not restartable. To fix this the whole event client state is resetted when ec_deregister() is called. Also a problem has been fixed that could cause two event client instances registered at qmaster depending how event client interface calls were used. Review: JG JG-2002-02-28-0 Enhancem.: Integrated multi level profiling into scheduler. Review: EB EB-2002-02-27-1 Enhancem.: Moved answer and feature functions into libsgeobjd.a. Added new object code managing host/hostgroup references. Added new test modules for range and hostgroup code. JG-2002-02-27-0 Enhancem.: Introduced multiple levels for profiling in sge_profiling. Bugfix: Scheduler did not get notified about global config changes. Review: EB EB-2002-02-27-0 Enhancem.: removed dependencies between libgdi and libsgeobj AH-2003-02-27-1 Bugfix: overriding queue limits with job request limits is incomplete for OS architecture dependent resource limits (NEC) Issue: 503 Review AS Changed sge_qmaster CR-2003-02-27-0 Testsuite: enabled bug test for issue 226 by allowing translation of macro MSG_CONFIG_CONF_GIDRANGELESSTHANNOTALLOWED_I in set_config() for SGE/EE 5.3 systems. AH-2003-02-24-0 Enhancem.: Documented MT safety of gdi Made more funcs MT safe Cleanup with gdi headers Review: CR CR-2003-02-26-2 Testsuite: testsuite sent more mails than allowed in Bugfix: enable_error_mails setup. CR-2003-02-26-1 Testsuite: fixed problem for missing MSG_COLONNOTALLOWED macro when testing SGE/EE 5.3. fixed bug in message macro translate() procedure CR-2003-02-26-0 Testsuite: automatic update of the message macro file when testsuite can't find specified macro. JG-2003-02-26-0 Bugfix: Classic spooling framework unnecessarily spooled array tasks in a certain situation where only spooling of the common part was required. Enhancem.: distinst installs new libraries if built as shared libs. Review: EB JG-2003-02-25-0 Enhancem.: Moved type handling commodity functions from mirror library to sgeobj library. Eliminated libspool's dependency on libmir. Review: EB AH-2003-02-24-0 Enhancem.: Documented MT safety of utilib made some funcs MT safe Review: JG JG-2003-02-24-1 Doc.: Updated module documentation to contain new libraries libsgeobj, libevc, libevm and libmir. Changed ADOC generation code in aimk to create individual documents for each library. CR-2003-02-24-1 Testsuite: - ignore second foreign system group at startup when no_main parameter is set - assume an extra entry in execd configuration for irix6 architecture (NICE values) CR-2003-02-24-0 Enhancem.: - Added commd profiling ( /tmp/commd/err.XXX file) - New sgecommdcntl options: -profile on|off|reset JG-2003-02-24-0 Enhancem.: Use spooling framework for qmaster spooling EB-2003-02-21-1 Cleanup: Moved functions from sgeobjlib to utilib AH-2003-02-21-0 Enhancem.: Initial step to JAPI job API that will be used for implementing DRMAA interface Review: AS EB-2003-02-21-0 Cleanup: replaced int by bool where applicable EB-2003-02-20-1 Enhancem. Enabled hostgroup code EB-2003-02-20-0 Enhancem. Added/Changed ADOC comments concerning sgeobj EB-2003-02-19-0 Testsuite: Added new test for issue #478 EB-2003-02-18-0 Testsuite: Added new test for issue #403 JG-2003-02-18-0 Cleanup: Created separate messages files for the new libs libevc, libevm and libmir. Updated ADOC docs for these libs. EB-2003-02-18-0 Testsuite: Added new test for issue #314 Review: CR EB-2003-02-18-0 Testsuite: Added missing test for issue #226 Review: CR JG-2003-02-17-0 Enhancem. Created new libraries: - libevc (event client code) - libevm (event master code) - libmir (mirror code) Issue: 466 Review: EB EB-2003-02-17-0 Bugfix: qhost -q -l arch=xx crashes if a grid execution host is down Issue: 492 Bugtraq: 4819479 Review: JG Changed: qhost JG-2003-02-14-0 Bugfix: Second migration of tightly integrated jobs failed if any host was used in both the first and second run of the job. Review: EB Changed: sge_qmaster CR-2003-02-14-3 Teststuie: comment from change by CR-2003-02-14-2 caused Bugfix: tcl error. CR-2003-02-14-2 Testsuite: fixed bug in get_ps_info() which caused qrsh test to fail Bugfix: CR-2003-02-14-1 Testsuite: throughput test now sets maxujobs parameter in scheduler configuration EB-2003-02-14-1 Testsuite: new test which will find #226 Issue: 364 CR-2003-02-14-0 Bugfix: changed not unique message id JG-2003-02-13-1 Bugfix: Fixed broken build for aimk -shared-libs for all architectures except aix EB-2003-02-13-0 Enhancem.: Merged code form Branch Resolved conflict libgdi <-> ligsgeobj CR-2003-02-13-1 Testsuite: Bugfix in read_array_from_file_data() - error reading in file without comment lines Enhancem.: New command line parameter: execute_func - execute a single tcl procedure CR-2003-02-13-0 Testsuite: shartree test now uses SHARETREE_RESERVED_USAGE=true JG-2003-02-13-0 Enhancem.: Improved ineffective implementation of removing entries in nonunique hashtables. Added a test program for cull_hash. Bugfix: Fixed improper use of hashtable in scheduler. Review: EB EB-2003-02-12-0 Enhancem.: created new library which contains CULL object related code. Several files were moved from libs/gdi into libs/sgeobj directory. CR-2003-02-12-0 Testsuite: - fixed tcl error (in test report generation) - updated scheduler profile string parsing for new string content (throughput test) CR-2003-02-11-0 Testsuite: bugtest for issue #140: Bugfix: using wait_for_jobend() call with job that is not mandatory running (using no_run_check option for this scenario) AA-2003-02-11-0 Bugfix: qmon crash fixed for pressing Why ? when several jobs are selected. Changed: qmon Bugtraq: 4816529 IZ: 490 CR-2003-02-10-1 Testsuite: - new SGE/EE version string for 5.3p3 - fixed bug for tar/zip package testing (could not enter package directory without zipped pgkadd files) CR-2003-02-10-0 Testsuite: - fixed tcl error for undefined warnings (caused from evtl. failed test) - increased timeout values for install_core_system test CR-2003-02-07-1 Bugfix: Compile warnings for ALINUX fixed CR-2003-02-07-0 Testsuite: Throughput performance test: Bugfix: - Error in Job to Project assignment. This causes wrong project job analysis. - enhancement for qevent client JG-2003-02-06-1 CVS: Created VMAIN_STABLE_2_TAG JG-2003-02-06-0 Bugfix: For tightly integrated parallel jobs, the handling of a task exit could be done multiple times resulting in too much accounting entries, probably wrong scheduling decisions and unnecessary spooling overhead. IZ: 438 Bugtraq: 4791238 Review: EB Changed: sge_qmaster CR-2003-02-05-4 Testsuite: Throughput performance test: Bugfix: Test submitted jobs without project switch JG-2003-02-05-0 Bugfix: path_aliasing truncated commandline of qrsh jobs Review: EB Changed: sge_execd CR-2003-02-05-3 Testsuite: Throughput performance test: gzip of data file to save disk space CR-2003-02-05-2 Testsuite: Throughput performance test: New protocol path for test results when local spooldirs are used on execds EB-2003-02-05-0: Bugfix: qstat -r shows wrong dependencies IZ: 478 Bugtraq: 4813188 Review: Has to be done Changed: qstat CR-2003-02-05-1 Testsuite: Throughput performance test: Event client works now for 5.3 Version CR-2003-02-05-0 Testsuite: Bugfix for Throughput performance test: tcl error JG-2003-02-04-0: Bugfix: qrsh silently truncated job arguments that exceeded 4k in size. Review: CR Changed: qrsh_starter CR-2003-02-04-0: Testsuite: - Throughput performance test: Event client works only for maintrunc -> added version check CR-2003-02-03-0: Testsuite: - Fixed Tcl error for unitialized array in run_tests() procedure (for warnings) - Fixed bug in issuezilla/141/check.exp (wrong wait_for_jobend parameter settings) AH-2003-01-31-1: Bugfix: wrong usage reported by qstat -j Bugtraq: 4794242 IZ: 446 Review: AS Changed: sge_execd CR-2003-01-31-3: Testsuite: 1) New procedure: mod_user() 2) test qconf for options -Auser and -Muser IZ: 91 CR-2003-01-31-2: Bugfix: qconf -Muser and qconf -Auser report no success messages Bugtraq: 4811230 Review: AH Changed: qconf CR-2003-01-31-0 Testsuite: Throughput performance test: - implemented resending of jobs in case of submit error CR-2003-01-30-0 Testsuite: Throughput performance test: - medium test: submit 8000 jobs, enable queues at 4000 pending jobs - use execd_params SHARETREE_RESERVED_USAGE=true CR-2003-01-29-0 Testsuite: Throughput performance test: Test connection to execd host before start to submit jobs (it was possible to lose the first job submits) JG-2003-01-29-0 Bugfix: startup failure of qrsh was reported as regular job exit. IZ: 380 Review: AH Changed: qrsh_starter, sge_shepherd, qsh JG-2003-01-28-0 Bugfix: qrsh crashed when job arguments exceeded 4k in size. IZ: 472 Bugtraq: 4807677 Review: CR Changed: qsh CR-2003-01-28-0 Testsuite: procedure compile_source(): Implemented loop for cvs update. Sometimes the cvs update failes due to an error. The testsuite will abort the compile procedure after three unsuccessfull update runs. AH-2003-01-24-0 Enhancem. uti: use get/set functions for acessing 'me' struct members uti: use get/set functions for localhost, fqdn_cmp, default_domain uti: made hash module reentrant partially made GDI reentrant made CULL reentrant made event_text() reentrant ensure sge_gdi_setup() is called once added MT support to RMON renewed commlib MT support CR-2003-01-24-1 Testsuite: bugfix: check issuezilla/126: - used global "host1" variable for getting hostname. This variable was initalized from migration and auto_reschedule test. - set load_thresholds and suspend_thresholds for the test queues to a high value - renamed "host1" and "host2" variable from migration and auto_reschedule test. CR-2003-01-24-0 Testsuite: - bugfix: get_tmp_file_name() - it was possible to get the same filename when calling procedure one after onother. - enhancm.: submit_job() - new parameter to disable output of job arguments - new test: issue_440 IZ: 460 AA-2003-01-24-0 Bugfix: qsub -C option analogue was missing in qmon remove #$ in submit dialogue's Prefix field to achieve same functionality as qsub -C before loading job with the file dialogue Review: JG Changed: qmon CR-2003-01-23-0 Testsuite: Bugfix for "check" command line option EB-2003-01-23-1 Bugfix: execd does process tracking even if no job has to be controlled. IZ: 416 Bugtraq: 4776016 Changed: sge_execd Review: JG EB-2003-01-23-0 Bugfix: qacct -l selection broken and qacct -nohist -l selection broken IZ: 455 Bugtraq: 4802171 Changed: qacct Review: JG AH-2003-01-23-1: Bugfix: qtcsh can't be used as normal tcsh Bugtraq: 4776821 IZ: 417 JG-2003-01-23-0 Bugfix: Set commlib timeouts in qsh analogous to qsub. Bugtraq: 4654993 SD-2003-01-22-0 Enhancem. Added code to source/libs/cull/cull_list.c to randomize the elements before calling qsort. This code is currently turned off until tests can be run to find out if it improves performance. AH-2003-01-22-1 Bugfix: STRING complex attribute handling with RELOP "!=" is broken IZ: 450 Bugtraq: 4805423 EB-2003-01-22-1 Bugfix: ":" in account string breaks accounting file IZ: 314 Bugtraq: 4713013 Changed: sge_qmaster, submit.1 Review: JG JG-2003-01-22-2 Bugfix: Empty sge_request file caused submission error. IZ: 393 Bugtraq: 4760981 JG-2003-01-22-1 Bugfix: qsub -C "" (do not parse special comments in script file) did not work. IZ: 449 Bugtraq: 480283 CR-2003-01-22-0 Testsuite: - throughput test enhancment: test sharetree with projects - procedure spool_array_to_file(): new parameter to switch of comment lines JG-2003-01-22-0 Enhancem.: Added timestamp to events. IZ: 457 EB-2003-01-22-0 Bugfix: qalter shows wrong priority number when using negative priorities with -p option IZ: #403 Bugtraq: #4769608 Review: CR Changed: sge_qmaster JG-2003-01-20-0 Bugfix: Fixed a problem in the event client interface when event clients did no regular fetching of events and had no busy handling configured. IZ: 445 JG-2003-01-16-0 Bugfix: Fixed a memory leak in the "simple scheduling interface" CR-2003-01-15-2 Testsuite: New paramter to run all tests in a subtree of the checktree: check PATH RUNLEVEL AA-2003-01-15-0 Bugfix: crash of qmon fixed: Cluster Config -> global -> Modify (Boingg) CR-2003-01-15-1 Testsuite: new performance test: throughput CR-2003-01-15-0 Testsuite: new procedure: create_gnuplot_xy_gif() - generate gnuplot xy charts (only if gnuplot is installed and in user binary path) CR-2003-01-14-3 Testsuite: modified remote procedures: - prototype for sendmail mail messages in mime format - new extra user dummy: ts_def_con2 for start_remote_prog() - bugfix in procedure resolve_upper_arch() CR-2003-01-14-2 Testsuite: modified file procedures: - get_tmp_file_name(): returned same file name when calling two times in a row. - convert_spool_file_to_html(): new option to return only html content without creating a file. - create_html_table(): new parameters to set border and align in tables new procedures: - read_array_from_file_data() - create_html_target() - create_html_image() JG-2003-01-14-0 Testsuite: Fixed and improved sharetree test. CR-2003-01-14-1 Testsuite: New command line parameter for testsuite: no_main2 : don't run main part and don't start setup ( used for sourcing this file) CR-2003-01-14-0 Bugfix: Fixed broken bild for glinux/alinux from CR-2003-01-13-0 CR-2003-01-13-1 Testsuite: qrsh test: set executable file permission for generated shell script and wait for nfs to share file. CR-2003-01-13-0 Enhancem. new qevent client. This event client is written for the testsuite to get job start and job end events. CR-2003-01-08-0 Testsuite: Using "ts_def_con" user for deleting spool Bugfix: spool directory with start_remote_tcl_prog() in install_core_system test. Install as no root user blocked deleting spool directory. JG-2003-01-08-1 Testsuite Testsuite only accepted branches as cvs revision, but rejected tags. JG-2003-01-08-0 Bugfix qtcsh did not build in a shared linked system (aimk -shared-libs) JG-2003-01-07-0 CVS: Created VMAIN_STABLE_1_TAG JG-2003-01-02-0 Enhancem. Prevent lex scanner to be generated after each checkout. Instead, the checked in scanner is used. SD-2002-12-19-2 Enhancem. A new qmaster_params attribute called SCHEDULER_TIMEOUT will allow the scheduler GDI event acknowledge timeout (default is 10 minutes) to be configured to a specific value. I have also added GEEE profiling information which reports the number of seconds required for specific parts of GEEE scheduling (pending job ticket calculation, job sorting, active job ticket calculation, and job placement). These changes will provide additional feedback for further reducing job scheduling overhead and improving scheduler scalability. AH-2002-12-19-1 Bugfix job arguments larger than 10k crash qmaster and introduced static dstrings IZ: 440 Bugtraq: 4792036 Review: EB JG-2002-12-19-0 Bugfix: - Accesstree was corrupted in the sge scheduler due to a bug in the event handling. - Wrong error output if a userlist shall be deleted that is still referenced in an execution host. JG-2002-12-18-0 Bugfix: Fixed problem with ACK mechanism between qmaster and execd. IZ: 352 Testsuite: Fixed problem retrieving local spool directory. CR-2002-12-18-0 Testsuite: Fixed bug which will result in a timeout error when testsuite tries to delete a local spool directory on a remote host JG-2002-12-17-0 Enhancem. Improved performance / scalability for job submit. JG-2002-12-16-0 Bugfix: qrsh without command reported "protocol error". IZ: 431 AH-2002-12-12-1 Bugfix conflicting policies can cause job being started and immediately suspended IZ: 437 Bugtraq: 4790592 Bugfix failover to shadow master leaves sge_schedd on the original qmaster host IZ: 441 Bugtraq: 4787623 Review: AS JG-2002-12-12-0 Cleanup: More general handling for code deleting from ranges in job handling. Removed unneccessary and expensive range handling in job_enroll(). Improved (example scheduler) simple_scheduler. EB-2002-12-11-1 Bugfix: Eleminated compilation warnings for architecture glinux in combination with gcc 3.2 EB-2002-12-11-0 Bugfix: job logging file exists but is empty in certain configurations IZ: 439 Bugtraq: 4791908 Review: AS JG-2002-12-10-0 Enhancem.: Create separate libraries for different spooling methods. Bugfix: Fixed broken build for creation of shared SGE libs. EB-2002-12-09-0 Bugfix: qstat -j shows messages concerning queues although schedd_job_info=false sge_schedd process consumes more memory than needed if schedd_job_info=true Job notification signals won't be delivered if user redefines suspend_method or terminate_method IZ: 432, 435, 436 Bugtraq: 4787598, 4790540, 4790547 Review: JG CR-2002-12-09-1 Testsuite: migration test now L10N compatible CR-2002-12-09-0 Testsuite: Removing evtl. existing access lists in install_core_system check CR-2002-12-06-2 Testsuite: Fixed bug in submit_del test ( running on system Bugfix: with large queue count will report errors on starup ) CR-2002-12-06-1 Testsuite: Fixed bug in access_lists test for sge systems Bugfix: CR-2002-12-06-0 Testsuite: - Setting shared library path in start_remote_prog() - Changed parameter settings for some tests JG-2002-12-05-0 Bugfix: Script submission for qrsh broke tight integration. CR-2002-12-05-1 Testsuite: fixed errors in error_states check and complex Bugfix: check CR-2002-12-05-0 Bugfix: Used wrong macro (SFQ) for fixing messages macros in commit CR-2002-12-03-1 CR-2002-12-04-0 Testsuite: Running more than one testsuite on the same host Bugfix: causes problems, because both testsuites will use same files in local /tmp directory - new procedures: get_tmp_directory_name() and get_tmp_file_name() Issue: 381 EB-2002-12-04-1 Testsuite: Fixed bug in checktree/functional/binary_submission CR-2002-12-03-1 L10N: Modified messages for spool files. The "#" (comment start) sign should not belong to l10n message. CR-2002-12-03-0 Testsuite: a) Bugfix: procedure del_access_list() b) Added new error handling for submit_job() (error open file) The error handling is used by binary_submission test. CR-2002-12-02-2 Testsuite: a) test job_environment: report only warning for JOB_NAME bug if version is <= 5.3p2 b) removed qrsh_trap test from qrsh test CR-2002-12-02-1 Testsuite: a) setting default configuration max_unheard to 00:01:00 b) new procedure: soft_execd_shutdown() CR-2002-12-02-0 Testsuite: a) Removed unused, duplicate message macro from source b) Removed wait_for_enter() call in submit_del test after pass2 (blocked automatic testsuite run) CR-2002-11-29-1 Testsuite: L10N bugfixes CR-2002-11-29-0 Testsuite: a) using testsuite root directory for lockfile, not Bugfix: testsuite results directory. b) get_version_info() may not reset install_core_system test if version is not ascertainable. CR-2002-11-28-2 Cleanup: un-commented message macros not used in source code L10N: CR-2002-11-28-1 Testsuite: added code to un-comment unused message macros ( code itself is also un-commented in procedure check_c_source_code_files_for_macros() ) CR-2002-11-27-3 Cleanup: a) removed/corrected duplicate message ids Testsuite: b) handle_vi_edit() procedure reports now an error if no expected parameter string matches the output of vi. CR-2002-11-27-2 Testsuite: removed hostname parameter in translate call. Using Bugfix: asterisk "*" instead. Now long hostnames will not L10N: influent string parsing anymore. CR-2002-11-27-1 Testsuite: a) submit_del check: debug modifications b) new procedure: check_c_source_code_files_for_macros() This procedure will return macros, that are unused in source code. c) open_remote_spawn_process() closing connection when execution script not exists on remote host. CR-2002-11-27-0 Testsuite: a) get_version_info() must test if product mode file Bugfix: exists, before opening the file. If file wasn't available a TCL error occured. b) increased default timeout for submit_job() and added more debug output for timeout errors. CR-2002-11-26-0 Testsuite: increased timeouts in qmaster complex check. Old timeout was too short to let the execd load values timeout at qmaster. CR-2002-11-25-1 Testsuite: a) start_remote_tcl_prog() in shadowd test broken, Bugfix: because of new testsuite configuration file setup (checked in CR-2002-11-20-0). b) project_access_lists test used wrong sge_macro() call for 5.3p2 system EB-2002-11-25-1 Bugfix: qstat -f output broken for pe jobs on same queue IZ: 430 Review: CR EB-2002-11-25-0 Bugfix: qsh -q hostname is not working IZ: 429 Review: JG CR-2002-11-25-0 Testsuite: test binary_submission -> added version check - check works only for sge/ee version > SGE(EE) 5.3p2 CR-2002-11-22-3 Testsuite: qmod - reschedule test: parallel job test should not submit job with -t 1 parameter (-t 1 is now a task job, this leaded into a missinterpretation of testsuite) CR-2002-11-22-2 Testsuite: qrsh output change in maintrunc - adapted changes into testsuite CR-2002-11-22-0 Bugfix: Uninitialized pointer caused segmentation fault in qsh/qrsh on submit only hosts. Testsuite: more info output for submit_hosts test CR-2002-11-21-1 Testsuite: new testsuite configuration parameter: add_compile_archs - new ts_config_version is 1.2 - Forced compilation for specified architectures CR-2002-11-21-0 Testsuite: remove source code macro dump file after source code update ( procedure compile_source() ) and create a new one. CR-2002-11-20-1 Testsuite: install_core_system check doesn't work without Bugfix: root password (noroot) CR-2002-11-20-0 Testsuite: Added l10n enhancements, new configuration files (new procedures for file spooling, gettext features, sge_macro (use macros from source code) ...) EB-2002-11-15-0 Testsuite: Added new tests which will find bug #126 IZ: 427 EB-2002-11-20-0 Bugfix: race condition if signals are to be delivered in job's startup phase IZ: 126 Bugtraq: 4780316 Review: AH CR-2002-11-19-3 L10N - modified message definitions - new command line parameters for infotext: raw : return localized format string __eoc__ : stop option parsing (after this option) JG-2002-11-19-0 Enhancem.: First part of a flat file spooling module containing generic functions for input/output of any objects in different formats. Cleanup: Several cleanup issues done in libuti and libgdi. AH-2002-11-18-1 Bugfix: Complex values for user defined complexes are rejected with global host. IZ 418 Bugtraq 4776754 Review AS EB-2002-11-15-0 Testsuite: Added two new tests which try to find bug #402 and #423 IZ: 425, 426 AA-2002-11-14-0 Kerberos: changes to fix build problems with krb5 EB-2002-11-14-0 Bugfix: - memopy leak in execd - stepsize 0 in array job specification results in qmaster exception - job env variables SGE_TASK_* might be overwritten by qsub -v IZ: 423, 424, 402, 264 Bugtraq: 4778757, 4778758, 4778762 Review: partially AH EB-2002-11-12-0 Enhancem.: Allow qsub -b y to handle built-in commands Issue: 25, 26 Review: AH JG-2002-11-04-0 Enhancem.: Added utility functions for profiling. Made scheduler use these profiling functions instead of former explicit implementation. Made event mirror and spooling test client use profiling functions. EB-2002-11-04-0 Enhancem.: Allow binary submission with qsub and allow script submission with qrsh Issue: 25, 26 Review: AH JG-2002-10-31-2 Bugfix: Event mirror interface: the event sgeE_JOB_MOD_SCHED_PRIORITY was not handled correctly. Spooling framework (classic): Removing local configurations used incorrect path. JG-2002-10-31-1 Bugfix: shepherd dumped core when starting command with unix behaviour (e.g. pe_start procedure). JG-2002-10-31-0 Enhancem. Documentation and some small enhancements for the spooling framework. Adapted naming conventions and wrote documentation for a number of functions in libgdi. JG-2002-10-29-0 Enhancem. Initial part of a spooling framework supporting the current "classic" spooling. A test program is provided mirroring all spooled data to a separate directory structure. Further documentation and fine tuning still to be done. JG-2002-10-24-0 Testsuite: qrsh tight integration test used unsupported qrsh -inherit option -N SD-2002-10-17-0 Bugfix: SGEEE - Fixed bug in the pending functional ticket calculation which incorrectly calculated the number of functional tickets for some pending jobs. Changed: sge_schedd AA-2002-10-16-0 BUGFIX: openssl0.9.6g in aimk.site SD-2002-10-15-1 Bugfix: Fix no newline problem with path aliases file If the last line in the path aliases file (sge_aliases) did not have a new line, qsub would core. IZ: 398 Changed: qsub, qsh, qmon, ... JG-2002-10-15-0 Bugfix: compilation failed using gcc-3.x IZ: 395 Enhancem. Introduced new module libs/gdi/sge_usage that contains usage and usage list related functions. IZ: 397 AA-2002-10-15-0 I18N: dist/locale/en/qmon_messages.ad and qmon_help.ad added JG-2002-10-14-0 Bugfix: Reduce communication overhead of sending orders and setting busy state of scheduler by using one sge_gdi_multi request. Removed misleading debug output in event client interface. IZ: 307 JG-2002-10-11-3 Enhancem. Replaced explicit event handling of sge_schedd by the use of the event mirror interface. IZ: 351 JG-2002-10-11-2 Bugfix: Fixes in event mirror interface: - events for queue (un)suspend on subordinate were not handled - fixed a potential memory leak JG-2002-10-11-1 Bugfix: Changing the event delivery interval of an event client had only effect for the next but one delivery. JG-2002-10-11-0 Bugfix: qlogin and qrsh checked the DISPLAY variable this is only necessary for qsh. AH-2002-10-10-1 Bugfix: Fix problem with user .cshrc error causing [pro|epi]log,pe-[start|stop] failure and setting queue into error state. [pro|epi]log,pe-[start|stop] are now started with "unix_behavior" i.e plain execvp(). This allows admins to prevent .cshrc/.profile from being sourced when #!/bin/csh -f or #!/bin/sh -f is used. IZ: 391 Bugtraq: 4756556 Review: AS Changed sge_shepherd JG-2002-10-08-2 Bugfix: An eventclient processing events slower than the event delivery interval could get into an endless loop. AH-2002-10-08-1 Bugfix: Commd read the host alias file each time when a hostname was not found in commd's host list. Now host alias file is read only at commd startup or when a COMMD_UNIQUEHOST request with refresh_aliases = true is received. Review: AS Changed: sge_commd JG-2002-10-08-1 Bugfix: Fixed memory access problem in event mirror interface. JG-2002-10-08-0 Bugfix: Fixed memory leaks and uninitialized pointer problems. AH-2002-10-07-1 Bugfix: Fixed problem with non-resvolvable hosts in host_aliases file causing wrong resolving results. IZ: 392 Bugtraq: 4756557 CR-2002-10-02-1 Testsuite: added sleep 1 command in procedure "delete_job" and Workaround increased wait timeout. Testsuite seems to have problems with deleting jobs. This may be connected with IZ issue #126 (qdel of just submitted job). When the signal delivering bug is fixed, the timeout and the sleep should be taken back. EB-2002-10-02-0 Cleanup: division by zero does not cause core IZ: 390 Changed: all binaries for solaris[64|86] architecture JG-2002-10-01-0 Enhancem. Initial Part of port for HP-UX 11 64 bit CR-2002-10-01-0 Bugfix: got compiler warning "unused variable `src_qname'" EB-2002-09-25-0 Bugfix: Exiting sgeee jobs will be shown in qmons dialog for pending jobs. IZ: 388 Changed: qmon EB-2002-09-30-0 Bugfix: Fix of EB-2002-09-12-1. JG-2002-09-30-0 Enhancem. Created a "simple scheduling interface" for the integration of the MAUI scheduler. source/libs/sched/simple_scheduler.c contains an example how to use it. IZ: 339 CR-2002-09-30-0 Testsuite: - qsh/qrsh does not accept -o/-e arguments anymore Bugfix: - added version support for 5.3.1beta9 and 5.3p2 EB-2002-09-25-1 Bugfix: Exiting sgeee jobs will be shown in qmons dialog for pending jobs. IZ: 388 Changed: qmon EB-2002-09-25-0 Bugfix: qconf gets commd timeout IZ: 308 Bugtraq: 4753669 Review: AH Changed: qconf CR-2002-09-25-0 Testsuite error when comparing host architectures and detecting duplicate compile hosts Bugfix: CR-2002-09-24-1 Testsuite Testsuite enhancement to test final packages Enhancem. IZ: 369 CR-2002-09-24-0 Bugfix: using SGE_STRUCT_STAT macro instead of "struct stat". Reason: compile error for SOLARIS86 CR-2002-09-23-3 Bugfix: 2 sentences need to be i18ned in master Bugtraq: 4746079 JG-2002-09-23-0 Enhancem. Created a gdi library interface (sge_mirror) for the mirroring of qmaster lists. IZ: 351 CR-2002-09-23-0 Bugfix: infotext doesn't support the reorder of conversion specifications (merge) Bugtraq: 4739811 Bug not fixed for ALPHA, ALPHA5, ALINUX architecture CR-2002-09-20-1 Bugfix: infotext doesn't support the reorder of conversion specifications (merge) Bugtraq: 4739811 SD-2002-09-19-0 Bugfix: Fixed restart_command for checkpoint jobs IZ: 383 Changed: sge_shepherd JG-2002-09-17-1 Bugfix: qsh jobs without or with local DISPLAY variable set have to be rejected. IZ: 342 Bugtraq: 4744523 Changed: sge_qmaster, sge_execd, qsh, submit.1 JG-2002-09-17-0 Bugfix: If a task of a tightly integrated job died through a signal (e.g. SIGXCPU), the job was not terminated and the signalling information not visible in the accounting record. Bugtraq: 4747829 IZ: 379 CR-2002-09-16-1: Bugfix: - removed message l10n for mail subjects CR-2002-09-13-1: Testsuite: - "-secure" now default aimk compile option Enhancem.: - try to get DISPLAY from user environment at startup - removed all "gets stdin" statements in expect/tcl scripts due expect "interact" problem. Using now a shell script for user input. This is a workaround for a "hanging" vi unter expect control. "gets stdin" seems to "block" the stdin for expect "interact" command - new menu entry: "(26) enter testsuite setup" - new menu entry: "(27) set command line options" (change testsuite options without exit testsuite;this enables the user to change setup without terminate testsuite -> existing rlogins stay open) (master and scheduler debug and setting SGE_DEBUG_LEVEL can now also changed "online") CR-2002-09-13-0: Bugfix: Typo in message Bugtraq: 4746705 EB-2002-09-12-2: Cleanup: Added two functions which might be used to retrieve the size of a CULL-list or element: cull_get_list_packsize() cull_get_elem_packsize() EB-2002-09-12-1: Bugfix: msg size used in tight PE integr. job start sync. protocol should be smaller Issue: 213 Bugtraq: 4658706 Review: AH Changed: sge_execd, sge_qmaster EB-2002-09-12-0 Bugfix: protocot doing termin. on failure for tightly integr. par. job could be leaner Issue: 214 Bugtraq: 4658716 Review: AH Changed: sge_qmaster JG-2002-09-12-1 Bugfix qrsh and qlogin accept unexpected options. qsh/qrsh/qlogin did not accept options -ac, -dc, -sc and -w IZ: 325 Bugtraq: 4716824 Changed: qsh JG-2002-09-12-0 Bugfix: qmake without information about parallel execution on commandline failed. qmake made an incorrect resource request if the ARCH environment variable was an empty string. IZ: 335, 327 Bugtraq: 4745399, 4745404 Changed: qmake EB-2002-09-11-0 Bugfix: schedd_job_info = true causes immense daemons memory growth IZ: 374 Bugtraq: 4742189 Review: AH Changed: sge_schedd EB-2002-09-10-0 Enhancem.: Initial Part of AIX 5.1 port Review: JG CR-2002-09-10-1 Bugfix: L10N - (setlocale() returns e.g. fr_FR on LINUX) cutting of characters after "_" for building message file directory path. CR-2002-09-09-2 I18N: invalid escape sequence in one of msgid Bugtraq: 4743505 Changed: inst_sgeee usage output AA-2002-09-05-0 I18N: - Calculation failure in Functional policy - Wrong mnemonic on qmon main window - qmon dies with changes in Edit Tickets on Solaris64 - suspend_threshold in queue icon display nearly not visible Bugtraq: 4742082, 4729085, 4740335 IZ: 350 EB-2002-09-04-0 Bugfix: qmod help output is incomplete Issue: 354 Bugtraq: 4741230 Review: AH Changed: qmod CR-2002-09-03-0 Testsuite: duplicate hostnames in configuration should be detected Issue: 300 also testing for duplicate architectures in compile_source procedure. AS-2002-09-02-0 Bugfix: Insufficient checking for PTF_*_PRIORITY values Issue: 333 Bugtraq: 4740407 Changed: sge_execd AH-2002-09-02-1 Doc: qmod man page for "-r" incorrect (user) daemons to crash at startup phase Issue: 193 CR-2002-09-02-1 Testsuite: - enhancements for project_access_lists test - added version information for 5.3.1beta5 - 5.3.1beta8 EB-2002-09-02-0 Bugfix: problems with destin_id_list syntax Issue: 313 Bugtraq: 4740350 Review: AH Changed: qsub, qalter, qsh AH-2002-08-30-1 Bugfix: solaris 32 bit: rlim_fd_max > 1024 can cause 32 bit daemons to crash at startup phase Issue: 371 Bugtraq: 4739596 Changed: sge_qmaster sge_execd sge_commd sge_schedd sge_shadowd CR-2002-08-29-1 Bugfix: corrected mistake in writing in debug messages for localization Changed: All binaries using i18n AA-2002-08-28-1 I18N: version string and copyright strings for infotext Review: AS Changed: infotext AA-2002-08-28-0 Bugfix: OK without hostname in host_configuration kill sge_qmaster Issues: 370 Bugtraq: 4732031 Review: CR Changed: sge_qmaster AH-2002-08-28-1 Doc: man page typo for sge_aliases Issue: 347 Doc: access_list(5) man page does not describe file format for SGEEE mode Issue: 357 Doc: Output of prolog/epilog for interactive jobs not mentioned Issue: 367 SD-2002-08-23-1 Enhancem. Updated MPI/Myrinet integration for latest release of MPICH-GM. Changed: startmpi.sh, sge_mpirun EB-2002-08-23-0 Bugfix: scheduler crashes if defaultuser has 0 shares in sharetree Issues: 363 Bugtraq: 4735972 Review: AS Changed: sge_schedd AA-2002-08-22-0 I18N: bunch of fixes for l10n Bugtraq: 4733802, 4733201, 4733089, 4733043, 4731976, 4731990, 4731967, 4731958, 4731944, 4731935, 4731273, 4729700 Issue: ????????, ????????, 375, ????????, ????????, ????????, ????????, ????????, ????????, ????????, ????????, ???????? EB-2002-08-22-0 Bugfix: Userset "defaultdepartment" accepts users in CLI Bugtraq: 4733859 Issues: 360 Review: AS Changed: sge_qmaster, qconf AS-2002-08-22-0 Bugfix: print correct progname in usage Bugtraq: 4735258 Changed: sge_qmaster sge_execd, sge_shadowd, sgecommdcntl EB-2002-08-21-0 Bugfix: qmaster error message during startup: global configuration doesn't exist - creating Bugtraq: 4665780 Issues: 226 Review: AS Changed: sge_qmaster, qconf, qmon AA-2002-08-21-0 Bugfix: gid_range bugfix for SGE mode in qmon Bugtraq: 4731288 Issues: 355 Changed: qmon AS-2002-08-19-0 Bugfix: gethostbyaddr binary not working on solaris64 Bugtraq: 4732545 Issue: 359 Changed: gethostbyaddr SD-2002-08-14-0 Enhancem. Updated Kerberos integration to handle cross-domain authentication correctly when storing the forwarded credentials in the credentials cache. Changed: put_cred JG-2002-08-08-0 Enhancem. Added RFE document for a mirroring module based on the event client interface. EB-2002-08-08-0 I18N: Messages in some types of mail notification are not i18ned. Bugtraq: 4728217 Changed: sge_execd, sge_qmaster EB-2002-08-07-0 Bugfix: maxujobs prevents dispatching even if job limit is not reached Issue: 345 Bugtraq: 4727515 Review: AH, AS Changed: sge_schedd CR-2002-08-02-0: Testsuite: delete_job: when wait_for_end is specified, delete job a second time CR-2002-08-01-0: Testsuite: - Several changes for maintrunc compatibility - better setup error behaviour Bugfix: ckpt_migration_test fails due to reference checks Issue: 340 JG-2002-08-01-0: Bugfix: Systems with security framework enabled did not work with cull version checking (JG-2002-07-17-0). CR-2002-07-31-1: Bugfix: English messages show up when running install in localized environment (fix for "fr" Solaris 8 and 9) Bugtraq: 4719203 Changed: all binaries JG-2002-07-29-0: Enhancem. Complete implementation and use of boolean and object CULL data types. Added module tests for CULL list functionality and CULL packing. Issue: 324 Cleanup: Removed unused function lSortLlist2. Bugfix: Fixed Bug in lDumpElem. Fixed memory access bug. CR-2002-07-31-0 Bugfix: addend missing #ifdef __SGE_NO_USERMAPPING__ lines to preserve sge_options[] array AH-2002-07-29-2 Doc: Describe the 'all' keyword for queue_list in sge_pe(5) and checkpoint(5). Issue: 319 Doc: Describe qconf -mqattr switch as the deprecated form of the -rattr and not of the -mattr switch. Mention that ?attr switch familiy operates also on PE and ckpt objects. Issue: 321 CR-2002-07-29-3: Bugfix: CLI: invalid option "-jid jid" for qconf in qconf -help Bugtraq: 4722060 (merged CR-2002-07-29-0 and CR-2002-07-29-1) EB-2002-07-26-1: Cleanup: - moved code - added ADOC comments - reimplemented sge_setenv() - eliminated some header files CR-2002-07-26-4: Bugfix: merged l10n fixes from V53_beta2_BRANCH CR-2002-07-26-3: Bugfix: removed debug output from qsh client EB-2002-07-26-0: Cleanup: removed duplicated code. EB-2002-07-24-0: Cleanup: removed code which someone copied in the past. CR-2002-07-24-0: Bugfix: wrong port output in qstat error message when qmaster Issue: 331 Bugtraq: 4719755 CR-2002-07-23-0: Testsuite: new feature for delete_job: wait for jobend Bugfix: project_access_lists and access_lists could not always delete parallel environments, because they was used not full completed jobs JG-2002-07-23-0: Enhancem.: CULL objects and CULL lists "know" which of their attributes have changed. EB-2002-07-23-1: Bugfix: job array task index values range not documented Issue: 283 EB-2002-07-23-0: Bugfix: qsub/qalter -l ... might select wrong resource. Issue: 328 Bugtraq: 4718880 Changed: sge_qmaster CR-2002-07-22-4: Bugfix: (merge) Junk messages in installation Bugtraq: #4705066 CR-2002-07-22-3: Testsuite: (merge) New version strings for 5.3.1beta3 and beta4 Enhancem.: EB-2002-07-22-2: Testsuite: testsuite didn't find aimk use -z 9 for cvs update EB-2002-07-22-1: Enhancem.: job input file Issue: #294 EB-2002-07-22-0: Bugfix: missing \n in qstat -j output CR-2002-07-19-0: Bugfix: fixed broken build from CR-2002-07-18-4 on some platforms. JG-2002-07-19-0: Doc: updated qmake manpage to reflect usage of the ARCH environment variable. JG-2002-07-18-0: Bugfix: qsh mechanism passing options to xterm was broken. help output for qrsh and qlogin contained unsupported options. CR-2002-07-18-4: Bugfix i18n messages without _() macro will not get into the gridengine.pot file. Fixed not i10n messages. CR-2002-07-18-2: Bugfix (merge from V53_beta2_BRANCH) i18n aimk compilation for SOLARIS86 did not translate i18n code Bugfix: i18n wrong function call in infotext binary - l10n not working Bugtraq: #4715919 JG-2002-07-17-0: Enhancem. Include version information in the cull packbuffer and do version check before unpacking messages. Issue: #322 CR-2002-07-15-0: Testsuite: solved dependence on user environment variable HOST. Bugfix: EB-2002-07-11-0: Bugfix: qacct displays incorrect accounting information Issue: #314 Bugtraq: #4713013 Changed: qacct CR-2002-07-10-0: Testsuite: don't check LAST_HOST variable in job_environement check. environment variable not supported in this release AH-2002-07-10-2: Enhancem.: Further improved load alarm/suspend alarm diagnosis messages Changed: sge_schedd AH-2002-07-10-1: Bugfix: Fixed problems with global load values preventing job dispatching Changed: sge_qmaster Issue: #311 Bugtraq: #4712023 EB-2002-07-10-0: Bugfix: missing environment variables in job env. AA-2002-07-10??0: Security: validity of CA cert -> -days in InitCA() in inst_sge CR-2002-07-09-1: Testsuite: merge from V53_beta2_BRANCH Enhancem.: - new check "functional/job_environment" testing job environment variables (e.g. JOB_NAME) Issue #267 CR-2002-07-09-0: Enhancem.: merges from V53_beta2_BRANCH (CR-2002-07-09-0, CR-2002-07-08-0) JG-2002-07-05-1: Enhancem.: new bitfield implementation in utilib. JG-2002-07-05-0: Testsuite: system_tests/qmaster/error_states: has to wait for job to leave qmaster before deleting pe due to new reference checking (EB-2002-06-27-1). JG-2002-07-04-1: Cleanup: Changed the macros for cull object definition to provide easier and more flexible object type definitions. Issue: #305 EB-2002-07-04-0: Doc: renamed "job array" to "array job" in the manual pages JG-2002-07-04-0: Bugfix: non immediate interactive jobs (-now no) failed, for immediate qsub jobs (-now yes) the job script was not checked. AA-2002-07-04-0: Bugfix: Job dependency display in qmon Issue: #275 affects only qmon Bugtraq: #4706929 JG-2002-07-03-0: Bugfix: Deleting ckpt objects and pe objects could make qmaster dump core. AH-2002-07-02-3: Bugfix: Fix problem with flittering load adjustment Issue: #309 Bugtraq: #4710514 Changed: sge_schedd AH-2002-07-02-1: Bugfix: Added missing newline character to error msg. EB-2002-07-02-0: Cleanup: see EB-2002-06-30-0 SD-2002-07-01-0: Enhancem: Updated MPICH/Myrinet integration EB-2002-07-01-0: Cleanup: see EB-2002-06-30-0 CR-2002-07-01-0: Bugfix: Fixed wrong message format defines for L18N/I10N EB-2002-07-01-0: Bugfix: Fix of EB-2002-06-17-0 EB-2002-06-30-0: Cleanup: in gdi library - removed code which was duplicated in the past - moved functions in differnet modules - renamed several function - wrote ADOC comments - ... EB-2002-06-28-0: Bugfix: Fix of EB-2002-06-17-0 JG-2002-06-28-0: Testsuite: Test case for tightly integrated parallel jobs, basic version testing pure functionality and behaviour in massive parallel environments. Issue: #13 JG-2002-06-27-0: Bugfix: qrsh -inherit failed with error "COMMPROC ALREADY REGISTERED" Issue: #303 EB-2002-06-27-1: Bugfix: prevent deletion of still referenced objects - PE objects - CKPT objects Issue: #302 (partially) AH-2002-06-27-0: Enhancem.: Improved load alarm/suspend alarm diagnosis messages EB-2002-06-27-0: Enhancem.: array job parameters in job array env Issue: #264 CR-2002-06-26-1: Testsuite: - new default parameter for wait_for_jobpending() (Merged) (return if job is already started) Enhancem.: - changed timeout values for access_list and project_acces_lists test EB-2002-06-26-1: Bugfix: qdel / qdel -f produces no output if execd is down Issue: #238 EB-2002-06-26-0: Cleanup: replace all %s in messages by macros SFN and SFQ Issue: #252 CR-2002-06-25-0: Bugfix: script syntax error in exe host installation Issue: #297 Bugtraq: #4705893 CR-2002-06-24-0: Enhancem. I18N : added message ids, switched off per default later: use qmaster/execd parameter to enable error id output change "on the fly". EB-2002-06-24-0: Bugfix: host group CLI contains "DO NOT MODIFY .." comment Issue: #293 EB-2002-06-23-0: Cleanup: Moved/renamed functions Moved messages into correct msg_* files Added ADOC comments Issue: #278, #281, #282 EB-2002-06-17-0: Enhancem.: Now it is possible to address pe tasks with the spooling functions for jobs. CR-2002-06-17-3: Bugfix: removed old i18n setup code CR-2002-06-17-0: Bugfix: Don't produce infotext error message "Binary not compiled with gettext!" JG-2002-06-14-0: Bugfix: The job environment variable JOB_NAME was set incorrectly. Doc: Updated man page - it contained false information concerning JOB_NAME. Cleanup: Cleaned up the way how execd detects and handles interactive jobs. Issue: #255 Issue: #266 CR-2002-06-13-6: Testsuite: testing issue #140 also for pending jobs Enhancem.: (merged from V53_beta2_BRANCH) JG-2002-06-13-0: Enhancem. qtcsh: If a command is started in the background using qrsh, the qrsh option -nostdin is used to avoid qrsh being suspended. Issue: #287 Bugtraq: #4701640 AA-2002-06-13-0: Fix: introduced lWhatAll, removed obsolete header files from SGE.i perlgui proto working again CR-2002-06-13-2: Testsuite: merged qstat -l / -q /-pe / -U option tests from V53_beta2_BRANCH Issue: #286 CR-2002-06-13-1: Testsuite: add. bugfix for html output in compile_source() CR-2002-06-13-0: Testsuite: error in html output in compile_source() AH-2002-06-12-1: Bugfix: Prevent use of soft requests for consumables. This addresses problems related to IZ #223 / BT #4700286. However it does not introduce debitation for soft resource requests. CR-2002-06-12-0: Testsuite: cleaning up html directory when starting compiling CR-2002-06-11-1: Testsuite: new testsuite checks Issue: #140 Issue: #141 Issue: #171 EB-2002-06-11-0: Bugfix: qstat shows waiting jobs of other PEs Issue: #284 Bugtraq: #4699665 Changed: qstat CR-2002-06-11-0: Testsuite: compile_source(): add install errors to html page when install host is not in compile hosts AH-2002-06-11-0: Bugfix: Complex default value was not used for load thresholds in cases when no load value was available. Issue: #223 Bugtraq: #4683852 CR-2002-06-11-0: Testsuite: compile_source(): tcl error on connection timeout Bugfix: EB-2002-06-07-0: Bugfix: disallow "none" in load_formula and addition tests for the correctness Issue: #215 Bugtraq: #4673738 EB-2002-06-07-0: Cleanup: ADOC comments AH-2002-06-07-1: Bugfix: Prevent qalter -l res=val on running jobs for consumable resources to prevent corruption of consumable managment Issue: #251 Bugtraq: #4683852 JG-2002-06-07-0: Cleanup: Replaced %s in messages with macros SFN and SFQ for all libs. Fixed problems with multiply defined messages. Issue: #252 JG-2002-06-06-0: Bugfix: Building Grid Engine with shared libs was broken. AH-2002-06-05-1: Bugfix: Fix problem with notify mechanism causing actual termination/suspension signal being delayed endlessly Issue: #231 Bugtraq: #4697491 JG-2002-06-05-3: Cleanup: Replaced %s in messages with macros SFN and SFQ for utilib. Issue: #252 JG-2002-06-05-2: Bugfix: qsub allowed submission of binary files Issue: #247 Bugtraq: #4696768 JG-2002-06-05-1: Bugfix: Error handling when reading configuration at qmaster startup was incomplete and produced confusing error messages. Issue: #226 Bugtraq: #4665780 JG-2002-06-05-0: Bugfix: Load values on HP-UX 11 multiprocessor machines were incorrect. Issue: #246 SD-2002-06-03-0: Bugfix: Fixed GRD 5.1.9 to SGEEE 5.3 upgrade Issue: #273 JG-2002-05-31-1: Doc: Improved howto for qrsh with ssh. JG-2002-05-31-0: Enhancem.: rshd: new switch -i: ignore /etc/nologin Issue: #272 AH-2002-05-29-1: Bugfix: Fix bug allowing non-privileged users to submit jobs with priority higher than 0 Issue: #271 Bugtraq: #4692957 CR-2002-05-29-2: Bugfix: - uninitialized function pointer compiler warning JG-2002-05-29-2: Enhancem.: Allow specification of arguments to [rsh|rlogin|qlogin]_[daemon|command] Issue: #232 CR-2002-05-29-1: Testsuite: - docu of HTML procedures CR-2002-05-29-0: Testsuite: - added prototype functions for HTML generation Enhancem.: - new parameter for testsuite: html_dir - compile_source() procedure will generate HTML output in html_dir directory JG-2002-05-29-1: Cleanup: make dstring functions return a const char * instead of char * Issue: #265 JG-2002-05-29-0: Cleanup: use of library functions, typos, improved job_get_id_string() Issue: #258 CR-2002-05-28-0: Testsuite: test new qrsh options -h and -hold_jid Issue: #66 JG-2002-05-28-0: Cleanup: common function calls to create the active job directories in execd. Replace static buffers for building paths by dynamic strings (dstring). Handle KEEP_ACTIVE with restarted jobs. Issues: #254, #261 AH-2002-05-28-1: Bugfix: Fix messages printed when GDI client fails contacting qmaster JG-2002-05-27-0: Bugfix: qsh ignored -S in sge_request(5) files Issue: #250 Bugtraq: #4682966 AA-2002-05-27-0: I18N.: I18N/L10N changes CR-2002-05-24-1: Enhancem.: merged L10N Changes for install scripts from V53_beta2_BRANCH CR-2002-05-24-0: Bugfix: - removed unused variable in infotext.c JG-2002-05-24-0: Bugfix: qalter -verify failed. CR-2002-05-23-0: Enhancem.: - added infotext_msg_parse.tcl - merge of infotext changes in V53_beta2_BRANCH JG-2002-05-23-1: Bugfix: qlogin jobs made execd crash. JG-2002-05-23-0: Cleanup: Made behaviour of sge_dstring_sprintf conformable to sprintf. New functions sge_dstring_sprintf_append (old behaviour of sge_dstring_sprintf) and sge_dstring_clear. JG-2002-05-22-1: Cleanup: Use a common function call to build paths in a jobs active_jobs directory. Use a common function call to output job/jatask/petask ids. Issues: #254, #258 JG-2002-05-22-0: Doc: Fixed description of the environment variable SGE_TASK_ID for non array jobs in submit man page. JG-2002-05-21-1: Bugfix: Recycling of additional group ids did not respect tightly integrated pe tasks. Issue: #263 CR-2002-05-21-0: Bugfix: smaller bugfixes in infotext binary JG-2002-05-21-0: Cleanup: Moved job_report related commodity functions to libs/gdi. Common code to create paths. Sum up usage of finished pe tasks in a container pe task and remove them from the ja tasks pe task list (merge of JG-2002-03-20-0 from V53_beta2_BRANCH). Other minor changes in the handling of pe tasks. Doc: Documentation of the pe task and pe task request types. Issues: 201, 254 EB-2002-05-21-1: Cleanup: Added missing ADOC comments in utilib EB-2002-05-21-0: Enhancem.: Prevent master from accepting jobs without id range specification Issue: #260 CR-2002-05-16-0: Enhancem.: - i18n changes for inst_sge (part 1) - renamed sge_printf into infotext (INteractive Formatted lOcalized TEXT) - added infotext features (-wait -auto -ask options) EB-2002-05-06-0: Bugfix: qhost -j is broken Issue: #259 Bugtraq: #4686157 Changed: qhost CR-2002-05-14-0: Cleanup: - Using dstring in sge_printf - some dstring function enhancements - fixed bug in sge_dstring_append() JG-2002-05-14-2: Doc: Coding styleguide, format for special comments. JG-2002-05-14-1: Bugfix: MPI tight integration rsh wrapper contained shortcut syntax not executed correctly by all shells. JG-2002-05-14-0: Cleanup: Introduced an own pe task datatype in cull, that replaces the former use of a complete job object for each pe task. Fixed some bugs in the handling of tightly integrated parallel jobs. Issues: #201, #221, #225 CR-2002-05-14-0: Cleanup: Removed unused schedd variable in sge_c_gdi_add() AH-2002-05-13-1: Change: Do no longer flush events immediately after orders arrival to 'guarantee' scheduling is done not more often than specified with schedule_interval in sched_conf(5). Changed: sge_qmaster CR-2002-05-13-0: Enhancem.: new source files for sge_printf binary in utilbin EB-2002-05-13-0: Bugfix: qconf -se shows "reschedule_unknown_list" param. Issues: #248 EB-2002-05-09-0: Cleanup: libgdi: - Moved some functions into libuti - Put all 'answer' specific functions into a module - Pooled code which was copied in the past - Eliminated 'old' pe-range code. Range module is used instead. - Modified and added several ADOC comments EB-2002-05-08-0: Bugfix: Fix of EB-2002-05-04-0 CR-2002-05-06-0: Testsuite: Fixed testsuite dependencies AH-2002-05-06-1: Bugfix: Fix typo in simple.sh job example EB-2002-05-04-0: Cleanup: Job object: - moved some attributes into the environment - created new module to access VA_Type lists - summurized submit client code - added ADOC comments CR-2002-05-03-1: Testsuite: - removed rsh with rlogin calls Bugfix: - compile source has set wrong LIBPATH on AIX - new parameter for create_shell_script(), start_remote_prog(), open_remote_spawn_process() (source_settings_file parameter) - new procedure: get_submit_error(): resolve submit error value EB-2002-05-03-0: Bugfix: Fixed bug of JG-2002-04-29-0. JG-2002-05-03-0: Bugfix: Fixed messed up JB_Type from JG-2002-04-29-0. EB-2002-04-29-1: Bugfix: schedd bugs reported by Patrik Koch on dev mailing list (Mon, 15 Apr 2002) - Memory leak in scheduler Issues: #243 Bugtraq: #4676340 JG-2002-04-29-1: Enhancem.: Contribution of Guiseppe Palumbo: Make the ibm-loadsensor script work with AIX 5.1.0 AH-2002-04-29-1: Change: In the sched_conf(5) man page it was wrongly documented that the maxujobs setting is only in effect when the user_sort policy is activated and that it is not available at all in a SGEEE system. Furthermore it was wrongly documented that the maxujobs limitation is applied separately to each priority group. Issue: #249 Bugtraq: #4681454 CR-2002-04-29-1: Testsuite: - check if qmaster/scheduler are alive before starting any test - modified timeouts of some tests - added error information on some tests - new procedures: is_scheduler_alive() get_scheduler_pid() JG-2002-04-29-0: Cleanup: Job object: - removed unused attributes and corresponging code - moved JB_end_time to jatask object - reordered attributes to reflect logical relations EB-2002-04-29-0: Cleanup: removed old code/comments/debug code CR-2002-04-29-0: Testsuite: added sleep in compile_source() procedure to give NFS time to update remote directories when creating dependencies CR-2002-04-26-1: Testsuite: depend was done on cvs checkout host, not at first compile host Bugfix: CR-2002-04-26-0: Testsuite: merges from V53_beta2_BRANCH AH-2002-04-26-1: Bugfix: - queue nsuspend > 1 was not used to suspend multiple jobs in a queue at a time in case of a suspend threshold alarm. Issue: #239 Bugtraq: #4675410 - use of suspend thresholds and user sort policy could cause "inconsistent access tree data" schedd logging and break user sort policy Issue: #240 Bugtraq: #4675416 EB-2002-04-25-0: Enhancem.: Summarize answer messages for array jobs Issue: #209 (partially) JG-2002-04-22-2: Cleanup: Function hostcmp no longer needed in qmake integration (remote-sge.c). JG-2002-04-22-1: Enhancem.: Improved verbose output in qmake. Start tasks on localhost also per qrsh - slower, but tasks are identifiable in accounting. JG-2002-04-22-0: Bugfix: If more then 1 task of a parallel job was started in a queue, retrieving exit code of a task and signaling a task was unreliable as the TASK_ID variable was no longer exported to the tasks environment. Replaced the TASK_ID mechanism by an entry in the jobs config. Issue: #236 JG-2002-04-19-0: Bugfix: Interactive jobs of any user failed if root had already the maximum number of group id's. Enhancem.: More detailed description in error output Issue: #237 SD-2002-04-18-3: Feature: Updated distinst to install mpi/myrinet PE integration SD-2002-04-18-2: Feature: MPI MPICH/Myrinet parallel integration Bugfix: Fixed dist/mpi/rsh to handle hostname before or after options CR-2002-04-18-1: Testsuite: Bugfix: testsuite did no aimk -only-depend CR-2002-04-18-0: Testsuite: Bugfix: testsuite only updates the c source code Issue: #187 EB-2002-04-17-2: Bugfix: error in qmasters job deletion code Issue: #235 AA-2002-04-17-0: Bugfix: Broken build for aix correction in csp.html JG-2002-04-17-1: Enhancem.: New qconf options -Astree and -Mstree Issue: #133 EB-2002-04-17-1: Cleanup: summarized client code Issue: #153 CR-2002-04-17-0: Testsuite: Enhancem.: new file procedure: get_all_subdirectories() Bugfix: report error if cat failes in procedure get_file_content() JG-2002-04-17-0: Cleanup: multiple defines for same error message Issue: #203 EB-2002-04-17-0: Cleanup: multiply defined function Issue: #202 EB-2002-04-16-0: Cleanup: building range strings for enrolled tasks JG-2002-04-16-0: Enhancem.: In simulation mode (qmaster_param simulate_hosts=1) one qsub call can submit any number of identical jobs by setting the job context variable SGE_MULTI_SUBMIT to the number of jobs to be created. CR-2002-04-12-3: Testsuite: demo_config_dir files should not be used in default settings file Issue: #163 CR-2002-04-12-2: Testsuite: removed all rsh calls, testsuite will now only use rlogin. Issue: #180 Bugfix: fixed tcl error when using help option JG-2002-04-12-0: Doc: Added howto describing how to use ssh as transport mechanism for qrsh. Issue: #228 CR-2002-04-12-0: Bugfix: corrected spelling mistakes CR-2002-04-11-4: Testsuite: - only produce warning when re_init core system resets cluster/schedd configuration - removed DEMO category - new interactive menu points: (23) compile and pre install binaries (24) compile and pre install, then install and run not completed tests (25) set testsuite debug level CR-2002-04-11-2: Testsuite: Remove temp script files at startup of next check function instead of next check. This should reduce the number of files to delete at one pass. CR-2002-04-11-1: Testsuite: Some further modifications for CR-2002-04-11-0 EB-2002-04-11-0: Bugfix: Solaris processor sets / Migration script - fix: solaris psets won't be used - fix: checkpointed job migration on bad exit state - some cleanup within shepherd concerning ckpt/pset code - ADOC comments Issue: #229, #174 Bugtraq: #4668148 CR-2002-04-11-0: Testsuite: Added testsuite categories, modified framework Enhancem. JG-2002-04-11-0: Bugfix: In simulation mode, job reports were not flushed at (simulated) job end. EB-2002-04-10-1: Cleanup: Moved messages EB-2002-04-10-0: Cleanup: Cleanup within CULL library - ADOC comments for all nonstatic functions - minor modifications JG-2002-04-09-0: Enhancem. Further generalization of the event client interface: - sge_locate_scheduler -> sge_locate_event_client - reinit_schedd -> reinit_event_client - special handling at registration not only for scheduler but any special event client - make datatypes of enums and use these types instead of int or u_long32 in access functions - extend event client modify request to allow changing of the busy state, remove special handling for scheduler busy state in qmaster - updated documentation JG-2002-04-08-3: Bugfix: qrsh -h made debug output (a job object). JG-2002-04-08-2: Enhancem. Performance optimization for the processing of job reports. JG-2002-04-08-1: Merge: Merged JG-2002-03-26-1 from V53_beta2_BRANCH Bugfix: At startup qmaster removed pending messages too late (after daemonizing instead of after enrolling). In rare cases (starting schedd immediately after qmaster daemonized on a fast machine) this could produce an error and cause a delayed registration of the scheduler. JG-2002-04-08-0: Merge: Merged JG-2002-03-21-0 from V53_beta2_BRANCH Bugfix: prioritization of messages to qmaster could block reports from execd's to qmaster and make the system unusable. Set "no prioritization" as default, behaviour can be changed via env variable SGE_PRIORITY_TAGS. Enhancem. Performance enhancement for the deletion of large numbers of jobs. CR-2002-04-05-2: Bugfix: Fixed compiler warning in ptf.c JG-2002-04-05-3: Doc: Linked both versions of the event client interface documentation (5.3 and maintrunc) from the modules overview document. JG-2002-04-05-2: Doc: Updated HTML documentation for the event client interface generated from adoc comments. CR-2002-04-05-1: Testsuite: - reinserted rsh test on testsuite setup configuration - merged large_cluster test from V52_beat2_BRANCH JG-2002-04-05-1: Cleanup: removed meanwhile meaningless comments in qlogin_starter module. CR-2002-04-05-0: Testsuite: - more function parameters for logfile_wait() Enhancem. - new procedure: close_logfile_wait() - output changes - new procedure: get_schedd_messages_file() - new procedure: check_schedd_messages() JG-2002-04-05-0: Enhancem. Further cleanup and enhancement of the event client interface: - general implementation for flushing of events replaces special handling for scheduler in qmaster - a new "busy policy" replaces special handling for scheduler with global variables in qmaster - a clientdata field in the event client object is used for checking of orders sequence numbers instead of using a global variable - lots of additional documentation SD-2002-04-04-1: Feature: Merges from V53_beta2_sx6_BRANCH SX architecture updates - Support dynamic setting of timeslice values - Set execd to high priority to prevent swapping - Only set # of processes limit for job (not process) - Updated SX checkpoint scripts - Renamed "necsx5" architecture to "sx" - Set resource limits after setting the OS job id in builtin_starter.c so that job resource limits will work on the SX-6. Added support to control restart timeout delays using environment variables. CR-2002-04-04-0 Testsuite: Bugfix: test shartree: deleting sharetree before adding a new one. Bugfix: start_remote_prog(): allow shell build-in command "cd" CR-2002-04-03-0 Testsuite: - replaced rsh calls with start_remote_proc() and open_remote_spawn_process() calls - new procedures: wait_for_remote_file() -> wait for file to appear on remote host (should fix NFS problems) init_logfile_wait() logfile_wait() -> new tail mechanism for parsing logfiles washing_machine() get_qmaster_messages_file() get_execd_messages_file() -> get path to messages files - enhanced procedures: start_remote_prog() JG-2002-04-02-2 Feature Compile time switch -DCULL_NO_HASH to switch of hashing of cull lists - mainly usefull for performance comparison. JG-2002-04-02-1 Merge: merged JG-2002-04-02-0, JG-2002-03-28-2, JG-2002-03-28-1, JG-2002-03-28-0 from V53_beta2_BRANCH Testsuite: Fixed problems when creating HTML charts in large_cluster test. Testsuite: - sge_procedures: new functions shutdown_qmaster, shutdown_scheduler and startup_scheduler. - large_cluster test: shutdown scheduler before initializing new test to get time for first scheduling. Testsuite: Create HTML output after each run level. Testsuite: large_cluster test: submit jobs in parallel from multiple hosts. CR-2002-04-02-0 Testsuite: testsuite should not fail if writing a script file fails Issue: #186 CR-2002-03-28-1 Testsuite: - merges from V53_beta2_BRANCH CR-2002-03-28-0 Testsuite: - new menu point: check messages files of execds and qmaster - new tcl file procedure: get_file_content() - new sge procedure: check_execd_messages() - new sge procedure: check_qmaster_messages() - new sge procedure: check_messages_files() - added new version: "SGEEE pre6.0 (Maintrunk)" - added new version: "SGE pre6.0 (Maintrunk)" JG-2002-03-28-0 Merge: merged JG-2002-03-27-0 from V53_beta2_BRANCH Testsuite: large_cluster test: - fixed problems due to change of runlevel - output cluster utilization in test report AH-2002-03-27-1 Enhancm.: Changed Maintrunk GRM_GDI_VERSION to a very high value and renamed from "5.3beta2" this version to "pre6.0 (Maintrunc)" CR-2002-03-27-2 Merge: merged inst_sge from V53_beta2_BRANCH CR-2002-03-27-1 Bugfix: Fixed broken shadow daemon qmaster startup functionality merge from V53_beta2_BRANCH SD-2002-03-26-2 Bugfix: Fixed "missing node XXXXX in dst trees" problem where a temporary share tree node was being sent to qmaster AH-2002-03-26-1 Bugfix: Found and fixed infinite loop in commd in case host formerly resolved by gethostbyname(3) is no longer resolvable. Also fix commd option -dhr which was -drh before. Bugtraq: #4658193 Issue: #212 CR-2002-03-26-5 Bugfix: fixed broken aimk -secure build CR-2002-03-26-4 Testsuite: merge from V53_beta2_BRANCH Bugfix: CR-2002-03-26-3 Merge: merge from V53_beta2_BRANCH Bugfix: - wrong error handling if qmaster is down and user tries qdel -f - changed code formating Bugfix: qstat -f -> 8 x times logging: "... neither submit nor admin host" Bugfix: can't resolve hostname "%-.100s" from act_qmaster file message Bugfix: qacct output formatting Bugfix: changed error messages when master commd is not reachable. Bugfix: removed forgotten #if0 - #endif statement which chauses scheduler to abort() on testsuite's path_alias check. Bugfix: scheduler generates high CPU load for 1-2 minutes when sge_commd is killed. Issue: #197 JG-2002-03-26-0 Testsuite: Installation with many hosts did not work due to some buffer (commandline?) being too small. Pass hostnames to install_qmaster in a file. AH-2002-03-25-3 Bugfix: gethostbyname -aname was broken Optimiz.: Ensure global conf is read only one time by AH-2002-03-22-0 code AH-2002-03-25-2 Bugfix: Ensure shepherd mapping signals do not hit shepherd itself because of signal handler not installed when signals are released by sigprocmask(2) AH-2002-03-25-1 Bugfix: Retrieve ignore_fqdn and default_domain at client/daemon startup from global cluster config to overcome problems with hashing of hostnames in certain cases. JG-2002-03-25-0 Testsuite: Bugfix: In a cluster with very similar hostnames testsuite wrongly reused open remote connections. AA-2002-03-22-0 Merge: - override, functional policy dialogue sorting - qhost formating - sec_lib SGE_KEYFILE, SGE_CERTFILE EB-2002-03-21-2 Bugfix: remove old configuration (EB-2002-03-21-0) EB-2002-03-21-1 Enhancem.: sort output of qconf -sep EB-2002-03-21-0 Bugfix: local config not written as admin user at startup Issue: #198 CR-2002-03-20-1 Testsuite: sending "vt100" when client host is using tset for terminal settings. EB-2002-03-20-1 Bugfix: can't receive request: READ ERROR - replaced closeconnection(0) calls within recvfromcommd() through closeconnection(1) calls in case of read error - improved logging; additional messages may be found in the messages files of the daemons which might help if these errors occure again Issue: #176 (partially) CR-2002-03-20-0 merge from V53_beta2_BRANCH Bugfix: qselect/qstat -U is (partially) broken Issue: #191 AA-2002-03-19-2: Merge: merge from V53beta2 branch - AA-2002-03-19-1, AA-2002-03-18-2, AA-2002-03-18-1, AA-2002-03-18-0 AA-2002-03-19-1: Feature: - install doc for CSP security 2 AA-2002-03-19-0: Feature: - install doc for CSP security CR-2002-03-19-2 Bugfix: additional fix for CR-2002-03-19-0 checkin CR-2002-03-19-1 Testsuite: (Merge from V53_beta2_BRANCH) - stop install_core_system check if qmaster is not reachable after installing last execd - shutdown_core_system(): only try to use root access if password is available CR-2002-03-19-0 Bugfix: Fixed broken build for IRIX6 arch CR-2002-03-18-1 (merge from V53_beta2_BRANCH) Bugfix: Mail on suspend wasn't working Issue: #86 Bugfix: Wrong error message when service sge_commd is not available Issue: #158 CR-2002-03-18-0 Testsuite: - qdel check: (merge from V53_beta2_BRANCH) 1) clear all error states on test cleanup 2) leave test loop on job submission error - new version numbers: SGEEE_5.3beta2_2 and SGE_5.3beta2_2 EB-2002-03-18-0 Enhancem.: Modifications for redhat 7.2 (gcc 2.96) AH-2002-03-15-5 Bugfix: Ensure after execd installation the host needs to be only exec host but not admin host or submit host EB-2002-03-15-0: Bugfix: qmaster code accepted qsub -t task specification which included the id 0 => loop in schedd Issue: #192 AH-2002-03-15-3 Doc: Added information about qmod [-e|-s|..] queue_list syntax (wildcards). AH-2002-03-15-1: Bugfix: Changed scheme used for creation of id used for qrsh -inherit sub-tasks to fix accounting data leak with multiple runs of a par. application within a job. Also write id into accounting file. Bugtraq: #4447392 EB-2002-03-13-0: Bugfix: handling of faulty migrate script - improved logging - shepherd internal state will be reset. Further migrate requests will be initiated Issue: #146 (partially) JG-2002-03-13-0 Doc: Improved the qmake man page to reflect the necessity to use the options -v PATH and -cwd. JG-2002-03-11-1 Doc: - Updated adoc description: Some new categories, replacement of the --Introduction categories, an example how to document cull objects. - adoc: Allow entries beginning with '-' to be included in the function index. - Updated some autodoc headers to reflect changes described above. JG-2002-03-11-0 Cleanup: Moved event client interface (client code) from scheduler to the gdi library and removed all scheduler dependencies left from previous cleanup step. JG-2002-03-08-0 Testsuite: Make testsuite work for users that use a csh as login shell (open_remote_spawn_process). AA-2002-03-07-0: Bugfix: Addition of acl to project generated core dump due to uninitialized pointer EB-2002-03-07-1: Bugfix: Array Tasks, inconsistent handling of multiple task ranges Issue: #49 EB-2002-03-07-0: Bugfix: inconsistences between qstat/qmon output and internal state Issue: #179 EB-2002-03-06-1: Doc.: Added documentation for (auto)rescheduling - removed RESCHEDULE_UNKNOWN from sge_conf - added reschedule_unknown to sge_conf - added r to when attribute of checkpoint - changed qsub -m doc. - added doc. for qmod -r Issue: #23 EB-2002-03-06-0: ADoc: JB_Type, JAT_Type, RN_Type got ADOC Header JG-2002-03-06-0: Doc.: Enhanced description of execd_params SET_[SGE|COD|GRD]_ENV. Issue: #120 AH-2002-03-05-1: Doc.: Fixed description of seq_no queue parameter in queue_conf(5). Bugtraq: #4490117 CR-2002-03-05-0: Testsuite: - removed forgotten wait_for_enter calls() CR-2002-03-04-1: Testsuite: - sge/ee installation path (product root) should not be longer than 60 chars ( workaround, because output of ps -ef command is truncated on some machines ) - fixed error in path_alias test: automounter /tmp_dir/ problem ! - submit_del (signal) test did not clear queues in error state at cleanup - modified error output in shutdown_master_and_scheduler() procedure CR-2002-03-04-0: Bugfix: hostname resolving check was executed even when ignore_fqdn = true was set. EB-2002-03-04-2: Bugfix: 'qdel ' might cause qmaster core depending on the tasks history. Issue: #173 AH-2002-03-04-3: Enhancem.: Added PVM 3.4.4 Virtual Machine ID setting to startpvm.sh from Andrew Shewmaker AH-2002-03-04-0: Bugfix: A set of fixes on initiative from Patrik Koch EB-2002-03-04-1: Bugfix: 'qalter -h xxx' caused core if task of array did not exist. Issue: #171 EB-2002-03-04-0: Bugfix: schedd_runlog file was not written by a 'qconf -tsm' command Issue: #170 CR-2002-03-01-2: Bugfix: error in CR-2002-03-01-1 when showing only one long project/dep/... by using e.g. -P LONG_PROJECT_NAME - fixed CR-2002-03-01-1: Bugfix: qacct truncates project/department/... names to Enhancem.: x chars. -> calculate column with by searching for largest element Issue: #150 EB-2002-03-01-1: Bugfix: inconsistences between qstat and qmon output Issue: #148 CR-2002-03-01-0: Testsuite: - using source/dist/util/arch from source directory and not from product_root/util/arch. This prevents errors on testsuite first setup call. EB-2002-03-01-0: Bugfix: "scheduler tries to schedule xxx twice" logging in qmaster Issue: #167 CR-2002-03-01-0: Testsuite: - runlevel of large_cluster performance test set to 300++ ( day ) CR-2002-02-28-1: Testsuite: Enhancem.: - new test: system_tests/sheperd/submit_del testing sheperd signaling with sheperd wrapper script. Issue: #125 (Part 1) Bugfix: - fixed typo in system_tests/clients/qdel - removed unused var in system_tests/shadowd/migrate CR-2002-02-28-0: Testsuite: Bugfix: - wrong error message in wait_for_end_of_all_jobs() Enhancem. - submit_job() generates output for timeout and eof error Enhancem. - 2 new parameters for create_shell_script() JG-2002-02-27-0: Doc: Documented the qrsh -noshell option in the qrsh/submit man page. Issue: #132 EB-2002-02-26-1: Bugfix: Fixed one problem repoted bye Patrik Koch. - switched call order in job_delete_not_enrolled_ja_task() - added some ADOC comments for scheduler EB-2002-02-26-0: Enhancem: Some modifications for EB-2002-01-23-1 - return value 25 for qrsh, qsh, qlogin, qresub - modified man qubmit man page CR-2002-02-26-1: Testsuite: - changed name of product feature for secure mode from "secure" to "csp" - compile option will now automatically add "-secure" to CHECK_AIMK_COMPILE_OPTIONS when product feature is "csp" CR-2002-02-26-0: Feature: qmaster installation script will now check if sge_qmaster binary is compiled with aimk -secure when -csp option is used, otherwise installation of qmaster will fail. JG-2002-02-26-0: Bugfix: qlogin did no longer work for solaris exec hosts. Issue: #151 CR-2002-02-25-0: Testsuite: new test: qsub -hold_jid JOBNAME test Issue: #124 AA-2002-02-25-0: Testsuite: security test changes EB-2002-02-25-0: Bugfix: missing header file AA-2002-02-25-0: Bugfix: missing header file, obsolete code removed EB-2002-02-24-0: Cleanup: in 'libuti' and 'libgdi' - moved functions from libgdi into libuti - rearranged/modified/created modules - some function names have been changed - added ADOC comments for all not commented non-static functions SD-2002-02-22-2: Feature: Added support to control restart timeout delays using environment variables. AA-2002-02-22-1: Bugfix: sec_lib.c memleak fix, reconnect functionality AA-2002-02-22-0: Cleanup: insure compile for tcsh LINUX6 crypto lib as libcrypto.a.insure CR-2002-02-21-0: Testsuite: added scripts/qlogin.sh (used for qmod reschedule test) CR-2002-02-20-0: Testsuite: cleanup of qmod reschedule test Issue: #118 EB-2002-02-20-0: Enhancem.: need means for limiting total number of jobs in a cluster Issue: #136 Cleanup: utilib; added ADOC comments EB-2002-02-19-1: Cleanup: Moved/deleted some functiones (part 2/2) Added ADOC comments Issue: #152 EB-2002-02-19-0: Cleanup: Moved/deleted some functiones (part 1/2) Issue: #152 EB-2002-02-18-1: Cleanup: Integarte common/job.c functions into libs/gdi/sge_job_jatask.c Issue: #142 SD-2002-02-18-2: Bugfix: - Set resource limits after setting the OS job id in builtin_starter.c so that job resource limits will work on the SX-6. CR-2002-02-18-1: Testsuite: added new check_function in qdel test EB-2002-02-18-0: Bugfix: Attempt to unlink directory + bogus error message - finished jobs have no task instanced => do not create a directory => put whole zombie job into one file $SGE_ROOT/default/spool/qmaster/zombies/ ... ... xx/yyyy/zzzz (zzzz is a file) Issue: #149 CR-2002-02-18-0: Testsuite: - alleviate testsuite setup CR-2002-02-15-0: Testsuite: - compile option: handle eof and response timeout CR-2002-02-14-0: Testsuite: - Bugfix: error on updating defaults file version - is_job_running() now first will get the current job state and then evtl. make a sleep (timing) EB-2002-02-14-0: Bugfix: - Error in qdel return message if task does not exist Issue: #144 CR-2002-02-13-0: Testsuite: secure feature: - add user keys for submit only hosts - submit large script mod. timeout for job end waiting - submit hosts test: wait for end of transfer before deleting job AS-2002-02-13-2: Bugfix: install "UPGRADE" file and shared libraries with "mk_dist" and "distinst" AS-2002-02-13-1: Enhancem.: added README.inst_sgeee to SGE_ROOT to explain how to install SGE as SGEEE AS-2002-02-13-0: Change: set version string to "5.3beta2_1" EB-2002-02-12-0: Bugfix: - step size was ignored from qdel command for job arrays - qmaster crashed when job array task id range comprises non existing task at the end Issue: #138 #139 AS-2002-02-12-0: Bugfix: Sun HPC installation script failed since default queue type now is already PARALLEL CR-2002-02-05-0: Testsuite: - wait_for_end_of_transfer() is now waiting that all lines of the qstat output have the same job state - install core system test using get_binary_path() for tar binary (used for secure product feature) - bugfix setting defaults file path (if not absolute) - rework of migration test (removed unnecessary error checks) JG-2002-02-05-1: Testsuite: Enable testsuite to setup a cluster on a filesystem where root has no write access. This is only a first step. A comprehensive solution has to be implemented, see issue #135 JG-2002-02-05-0: Testsuite: Enhanced "large_cluster" test: - testcase "throughput cluster" (loadcorrection and loadthresholds switched off) - testcase "usersort" (scheduling param user_sort enabled) - monitor size of qmaster and scheduler - check if all slots are filled in first scheduling run AS-2002-02-04-1: CVS Created V53_beta2_BRANCH ---------------------------------------------------- AS-2002-02-04-0: Doc: updated courtesy download pages to reflect changes for Grid Engine 5.3beta2 download CR-2002-02-04-0: Enhancem.: aimk -DNO_USER_MAPPING_AMBIGUOUS_IN_CHECK will disable the user mapping check for incoming mapping requests. Issue: #129 JG-2002-01-30-2: Enhancem.: Higher precision for scheduler profiling. JG-2002-01-30-1: Doc: Updated qconf manpage to document options -secl and -kec (see JG-2001-11-26-0). Issue: #100 JG-2002-01-30-0: Doc: Updated qconf manpage to document options -Auser and -Muser (see JG-2001-11-08-1). Issue: #90 CR-2002-01-29-2: Testsuite: create foreign user keys for secure system CR-2002-01-29-1: Bugfix: broken aimk -umap build Issue: #127 JG-2002-01-29-1: Doc: Updated qconf manpage to reflect changes from JG-2002-01-10-0. Issue: #115 JG-2002-01-29-0: Testsuite: Enhanced "large_cluster" test: - check product type (sge/sgeee) - create HTML report with charts generated by gnuplot EB-2002-01-29-0: Enhancem.: Improved logging - product mode string will be printed during startup and shutdown CR-2002-01-29-0: Testsuite: - new settings file version "6" (new parameter product_feature) Enhancem.: - added support for secure SGE/EE system - removed command line parameter test_perm - added command line parameter reset_conf - updated changes from install script - testing product feature file on startup AH-2002-01-29-0: Bugfix: - scheduler had problems with iterating over PE slots range from max to min in case of $pe_slots alloc rule AA-2002-01-28-2: Feature: - sge_ca -usercert added to generate keys/certs for users in file EB-2002-01-28-0: Bugfix: - "qselect -l h=" prints error message. AA-2002-01-28-1: Bugfix: - qmon unable to contact master message fix - func proto show_info_job,.. AA-2002-01-28-0: Bugfix: - GCC_NO_STRICT_PROT broke build on alinux GCC_NO_STRICT_PROT -> GCC_NOSP in aimk and security/sec/Makefile - memleak fixes in security/sec/sec_lib.c AS-2002-01-28-0: Change: removed SECURE defined from CULL headers" SD-2002-01-25-2: Bugfix: Changed aimk variable name GCC_NO_STRICT_PROTO to GCC_NO_STRICT_PROT variable because /bin/csh on the SX-6 only allows 18 character variable names. AH-2002-01-25-1: Bugfix: Removed last reference of no longer existing QSI from man pages. The QSI functionality is replaced by {starter,terminate,suspend, resume}_method functionality Enhancem.: Default queue type is now batch, interactive and parallel. EB-2002-01-25-1: Bugfix: - shadowd did not remove the lock file if it encountered that there may be another master. Enhancem.: - A shadowd which tries to start a new master has to check whether the old one is still enrolled at the old master commd. If this is the case the shadowd has to wait for another cycle. AH-2002-01-25-0: Enhancem.: - fixed man page doc issues #31, #84, #85 * changes on -hold_jid accepting both jobid and jobname * default location for Grid Engine host aliases file not descr * qstat(1) does not explain meaning of 'd' state indicator CR-2002-01-25-0: Bugfix: changed some settings to prevent timeouts EB-2002-01-25-0: Enhancem.: new schedd_param: POLICY_HIERARCHY JG-2002-01-25-0: Enhancem.: If killing a qrsh job (by the pid of the qrsh_starter) is not successful (because the started application left the process group) a second kill request will kill the process group of the shepherds child (the rshd) to ensure the job is terminated. AH-2002-01-24-0: Enhancem.: - added 'maxvmem' as new per job usage information in qmon and accounting - fix problem with wrong 'maxvmem' information in job finish mail for non-pdc architectures - ensure daemons log 'starting up'/'controlled shutdown' messages independently of the configured log_level SD-2002-01-24-4: Bugfix: IRIX C compiler didn't like default: followed directly by a } in a switch statement. AA-2002-01-24-3: Bugfix: coshepherd Makefile SLIBS was missing EB-2002-01-24-2: Bugfix: Typo: NO_REPRIORIZATION -> NO_REPRIORITIZATION AA-2002-01-24-2: Feature.: sge_ca permission problem for daemon cert fixed AA-2002-01-24-0: Feature.: sge_ca -copy to install user cert and key CR-2002-01-24-0: Enhancem.: updated install script (inst_sge) changes JG-2002-01-24-3: Bugfix: Scheduler profiling could not be switched off at runtime once it was enabled. EB-2002-01-24-1: Bugfix: Give other clients (gdi requests) the chance to be handled by the commd before we send a next request of a fragmented qdel request. JG-2002-01-24-2: Bugfix: np_load_avg was not correctly computed for multi processor machines. EB-2002-01-24-0: Bugfix: Fixed "bug" within the code handling task id ranges which caused performance problem in scheduler for huge fragmented array jobs and array jobs with a step size != 1. AS-2002-01-24-2: Enhancem. added SUN MPI install support for 'distinst'" AS-2002-01-24-1: Enhancem. Added PE 'make' template AS-2002-01-24-0: Enhancem. Added SUNHPC loose integration package AA-2002-01-24-0: Feature: Security: - sge_ca script changes: rand file generation check of dn - rand file search paths - GCC_NO_STRICT_PROTO for alinux JG-2002-01-24-1: Testsuite: New test "large_cluster" testing performance and scalability of a Grid Engine cluster. Simulating a large cluster using the simulation mechanisms provided by JG-2002-01-18-0. Tests performance dependent on - number of hosts and number of jobs - number of hosts and size of a sharetree (GEEE) JG-2002-01-24-0: Testsuite: New functions in sge_procedures: - add_exechost: add an exec host object - add_user : add a user (GEEE) - del_user : delete a user (GEEE) AS-2002-01-23-1: Enhancem. set execd priority not higher than PTF_MAX_PRIORITY SD-2002-01-23-3: Bugfix: Fixed setting of shared override tickets in SGEEE_CLASSIC_SCHEDULING mode. EB-2002-01-23-1: Enhancem.: max_u_jobs parameter to limit the maximum number of jobs for a user. EB-2002-01-23-0: Bugfix: qsub -e/o caused crash of qmaster SD-2002-01-23-2: Bugfix: Initialize rusage structure to zeroes AS-2002-01-23-1: Enhancem. set execd priority not higher than PTF_MAX_PRIORITY AA-2002-01-23-1: Feature: Security: - fallback key/cert paths - removed gsd.refresh_time debug message CR-2002-01-23-1: Testsuite: - new procedure: close_open_rlogin_sessions() Enhancem.: - closing all rlogin sessions on TCL error - send CTRL + C on each close_spawn_process() call to stop evtl. running applications at remote host CR-2002-01-23-0: Testsuite: - updated resolve_version for SGE(EE) 5.3beta2 - bugfix for SGE installation (install_core_system) AA-2002-01-23-0: Feature: Security: - change of cert/key paths - sge_security_ck_to_do() added kerberos and CSP sec funcs now in module libs/gdi/sge_security.c - clearance of connectionlist changed CR-2002-01-22-0: Testsuite: - install_core_system check will now use default values for most all cluster configuration setup and report configuration changes through sge_inst script - re_init cluster after tcl_error will now restore cluster configuration for most parameters AS-2002-01-21-2: Feature: updated version and GDI version AS-2002-01-21-0: Bugfix: fixed typo in scripts/distinst JG-2002-01-22-0: Bugfix: Tight integration using the mpi scripts: starting a task from within a task failed, as the environment variable TASK_ID was exported to the second level qrsh call and qrsh tried to allocate this TASK_ID a second time. CR-2002-01-21-1: Testsuite: - setup checking length of product_root_directory - setting default load_thresholds to "np_load_avg=11.00" - Bugfix in submit hosts test - Bugfix for sharetree test (no timeout behavior for CLASSIC_SGEEE_SCHEDULING case) CR-2002-01-21-0: Testsuite: - changed menu point 9 ( run all tests at a special starttime) Enhancem.: SD-2002-01-18-2: Enhancem. Minor PTF fix for setting nice values on NEC SX-6 EB-2002-01-18-1: Bugfix: - qalter of job name caused core of master JG-2002-01-18-0: Enhancem. Simulation mode for large clusters. When the qmaster_param simulate_host=1 is set, a cluster with a high number of hosts can be simulated. EB-2002-01-18-0: Bugfix: - reactivated qdel output for already deleted jobs - enabled signalling for already deleted jobs CR-2002-01-18-1: Testsuite: - general performance modified cluster configuration Bugfix: without restoring it. CR-2002-01-18-0: Testsuite: - new command line parameter: re_init_on_tcl_error Enhancem.: (if set the testsuite will do a install re_init before starting the next test) - extend timeout values in submit_hosts check AA-2002-01-17-1: Bugfix: - qmonMirrorMultiAnswer - I18N in qmon CR-2002-01-17-1: Testsuite: - submit_job(): Enhancem.: new error state -13 for unkown option error Bugfix: - fixed broken file_parsing test - source of tcl files at startup of each test CR-2002-01-17-0: Bugfix: ERROR(()) call without string limitation "SFQ" in execd caused segmentation fault when job name length > (ca.) 2K byte EB-2002-01-17-0: Enhancem.: Improved logging of shadowd AA-2002-01-17-0: Feature: Security: - cert scripts JG-2002-01-16-0: Enhancem.: Execd can send load values for another host than itself and qmaster will accept these load values. Simulation mode to make qmaster not trash load values for simulated hosts can be switched on by qmaster_param simulate_hosts=1. Performance improvement: Execd deliveres info whether a variable is static or not. CR-2002-01-16-1: Testsuite: - fixed broken migration test Bugfix: - fixed broken job_state_handling test - new optional parameter for submit_job(): cd_dir: set working directory EB-2002-01-15-0: Enhancem.: per user namespace for job dep. names CR-2002-01-16-0: Testsuite: - faster lockfile check - new settings file version: 5 (new foreign users/group setup) - syntax check for testsuite users/group setup - set_exechost(): ignore unchangable values - submit_job(): new error states for access errors - get_qstat_j_info(): joining lines, containing no ":" in order to get complete scheduling info - new procedure: get_scheduling_info() - new procedure: add_access_list() - new procedure: del_access_list() - job_state_handling test: using new procedure get_scheduling_info() - qrsh test: update for defaults file changes - new test: geee/project_access_lists - new test: functional/access_lists SD-2002-01-15-1: Enhancem.: NEC SX-6 compilation changes: Added missing strptime and fnmatch functions (from BSD) Added support for NECSX6 architecture (maps to NECSX5) Updated aimk and Makefiles A few minor compiler related changes to source code EB-2002-01-15-0: Enhancem.: 'qdel all' and 'qdel -uall' requests will be split up into multiple requests JG-2002-01-15-1: Bugfix: qconf -mattr only handled string attributes, changing attributes of other datatypes made qconf dump core. AA-2002-01-15-0: Feature: Security: - sge_security_verify_user wraps sec method specific funcs sec_verify_user() and krb_verify_user() - addition of uniqueIdentifier to keep the Unix user name to certificate generation scripts/templates - xcheck between Unix user and certificate uniqueIdentifier field JG-2002-01-15-0: Testsuite: Fixed a problem of the submit_host check: Replaced "hostname" with "uname -n" as hostname is not found on some architectures (e.g. on irix6 hostname is located in /usr/bsd/bin). JG-2002-01-14-1: Bugfix: Removed duplicated code that caused the Insure++ debugger to (falsely) detect a READ_DANGLING error. JG-2002-01-14-0: Enhancem.: execd_param SET_[SGE|COD|GRD]_ENV: If all types of job environments are configured to be off, the default SGE is used. CR-2002-01-11-0: Testsuite: wrong parsing for submit_job() produced error Bugfix: in qsub test. EB-2002-01-11-0: Bugfix: not existing shell in jobscript in combination with shell_start_mode=unix_behaviour set queue and not job into error state. AA-2002-01-11-0: Feature: Security: - scripts for generation/installation of certs, 1st draft under dist/util/sgeCA - changes to script/distinst, script/distinst.site JG-2002-01-11-0: Bugfix: Fixed a memory corruption bug (READ_DANGLING and FREE_DANGLING) in libcull. Added code to suppress warnings from the Insure++ debugger for some correct code sections. Make Insure++ output filenames and line numbers. SD-2002-01-10-2: Enhancem: Updated security modules aimk for linux CR-2002-01-10-1: Testsuite: a) new sge_procedures: - wait_for_job_state() - wait_for_queue_state() b) new test: - system_tests/qmaster/error_states SD-2002-01-10-1: Enhancem: Added SGEEE policy hierarchy support (not yet activated) Minor GSS security modules update Bugfix: Fixed uninitialized structure in sge_calc_tickets EB-2002-01-10-0: Bugfix: qrsh -notify CR-2002-01-10-0: Testsuite: Enhancem.: settings file parameter "user_environment" Testsuite will read users environment value if no var=value syntax is used. So if user_environment is set to DISPLAY the testsuite will export the VALUE of the DISPLAY variable on the local host to each remote host. JG-2002-01-10-1: Testsuite: Make the number of parallel rlogin sessions dependent on the limit for file handles. JG-2002-01-10-0: Enhancem.: Additional check of the host parameter to qconf -ke - only accepts non option parameter. EB-2002-01-09-0: Bugfix: qselect -pe Bugtraq: #4584673 SD-2002-01-09-1: Enhancem.: Added -clearusage info to qconf -help AA-2002-01-09-0: Enhancem.: openssl lib path, I18N Linux uses libc gettext -> changes in aimk JG-2002-01-09-0: Enhancem.: qconf -ke has to be called with a host(list) or the keyword "all". qconf -ke without hostlist no longer kills all hosts. Issue: #83 CR-2002-01-08-0: Testsuite: - New settings file version "4" because of new parameters "dns_domain_name" and "user_environment". - unsetting GRD_ROOT and CODINE_ROOT variables (if set) at startup - qmod/reschedule test is now testing more, but still not full implemented. - changed output of handle_vi_edit() - new parser procedure parse_qconf_se() - commented open_root_spawn_process() this procedure should not be used anymore. open_remote_spawn_process() should be used therefor. - new procedures: get_complex(), set_complex(), get_qconf_se_info() - enhanced sumbit_job() for specific error value handling - new check: system_tests/qmaster/complex JG-2002-01-08-0: Enhancem.: Made "with I18N" default for solaris, solaris64 and glinux architectures. Fixed broken build with I18N. AH-2002-01-08-0: Cleanup: Removed old and unused starter code in shepherd JG-2002-01-07-1: Testsuite: handle_vi_edit sometimes "worked too fast" (modified file had same timestamp as original). Added a sleep of one second. JG-2002-01-07-0: Bugfix: Fixed a memory leak in execd. Added code to suppress warnings from the Insure++ debugger for some correct code sections. AA-2001-12-21-0: Feature added sge_ca script RAND_load_file sec_setup_path path changes JG-2001-12-21-0: Enhancem. Eliminate warnings from the Insure++ compiler AA-2001-12-20-1: Feature: Security: - commdcntl skips sge_security_init() - unpack free of elem fix - removal of obsolete setup routines in seclib EB-2001-12-20-0: Bugfix: Hold tid list of job submission dialog honours tid range specification. Issue: None (Fix of Enhancem.: EB-2001-12-12-0) CR-2001-12-20-0: Testsuite: moved job start procedures from job_state_handling test to the global sge_procedurs.tcl file. (submit_waitjob_job(), submit_time_job(), submit_error_job()) EB-2001-12-19-0: Enhancem.: qmaster writes heartbeat-file during job deletion CR-2001-12-19-1: Testsuite: new test: job_state_handling - testing job states for array and "normal" jobs (hold, pending, running, finished, error, timehold jobhold, suspend is tested in qmod test) - testing max_aj_instances - testing max_aj_tasks new parser procedure: parse_qstat_j() (parse qstat -j) new procedure: get_qstat_j_info() (get qstat -j information) new procedure: get_job_state() (parse qstat -f for job/array states) Bugfix: unlock testsuite after fault tcl command code CR-2001-12-19-0: Bugfix: added missing ":" for qstat -j output for "job-array tasks" AA-2001-12-19-0: Bugfix: scrollbar anomalie in cluster config, made scrollbars static JG-2001-12-19-1: Bugfix: Limits were not correctly set on hp10 platform - nonetheless most limits will have no effect due to limitations of the hp10 platform. JG-2001-12-19-0: Bugfix: Limits did not work on glinux platform JG-2001-12-18-1: Testsuite: - Fixed qsub check_submit - now it submits jobs from any host to any host in the cluster as expected. - extended tcl library function "submit_job" by parameters "host" on which to execute qsub and "user" as account that will execute qsub. EB-2001-12-18-0: Bugfix: 'qstat -f' showed pending tasks multiple times Rescusciated '-g d' option of qstat Issue: None (Fix of Enhancem.: EB-2001-12-10-1) JG-2001-12-18-0: Testsuite: - new test procedures in qsub and in qrsh tests that test the starter_method. SD-2001-12-17-2: Enhancem.: - added verbose information to get_cred security module - updated DCE documentation CR-2001-12-17-1: Feature: - Grep for host-type and foreach loops, optimize via first/next hash table functions. - set exit state of qconf -ks to "1" if no scheduler is enrolled - is_requested() will accept CE_name AND CE_shortcut now. AH-2001-12-17-0: Bugfix: Ensure immediate return of sync send_message() in case receiver is not enrolled CR-2001-12-17-0: Testsuite: - handle_vi_edit procedure now uses now open_remote_spawn_id procedure - bugfix in setup_queues / install_core_system check - bugfix in performance/cluster/check.exp (hanging testsuite) JG-2001-12-17-2: Bugfix: Bugfix for JG-2001-12-17-0 handling "none" as starter_method. EB-2001-12-17-1: Bugfix: qstat -j did not show a monitoring message Issue: #113 Bugtraq: #4614501 JG-2001-12-17-1: Testsuite: Testsuite framework will output stack trace if a TCL error causes a test to fail. JG-2001-12-17-0: Feature: Implemented starter_method for qrsh. Issue: #6 EB-2001-12-17-0: Feature: Compatibility mode (sge_request, SGE_*) AH-2001-12-14-0: Bugfix: SGE's hostaliasing can prevent qmaster restart Issue: #95 Bugtraq: #4616534 AA-2001-12-14-1: Bugfix: got a X protocol error, fixed -> AA-2001-12-14-0: AA-2001-12-14-0: Bugfix: Sharetree did not show the correct user data after leaving and reentering the sharetree dialogue when the sharetree had been scrolled down JG-2001-12-14-0: Doc: ADOC documentation for the event client interface. HTML version in doc/devel, linked to module_doc. EB-2001-12-13-0: Bugfix: Qmaster automatically corrects invalid task id ranges during job submittion. Issue: None (Fix of Enhancem.: EB-2001-12-10-1) AA-2001-12-13-0: Feature: added max_aj_instances and max_aj_tasks to global cluster config qmonMirrorMulti -> qmonMirrorMultiAnswer EB-2001-12-12-0: Bugfix: Changed/enhanced qmon's "Job Control" and "Job Submission" dialog so that they respect unenrolled pending tasks. Issue: None (Fix of Enhancem.: EB-2001-12-10-1) JG-2001-12-12-0: Copyright: Create and link libXltree as shared lib; it is LGPL Code. EB-2001-12-11-0: Bugfix: Printing of zombie jobs in qmon and qstat Issue: None (Fix of Enhancem.: EB-2001-12-10-1) CR-2001-12-11-1: Testsuite: Start sgecommdcntl as root if root access is available CR-2001-12-11-0: Testsuite: Using catch{} construct to suppress tcl errors when creating a report message JG-2001-12-11-0: Bugfix: qrsh can not really ignore CTRL-Z (as intended by JG-2001-12-10-0). Pressing CTRL-Z in a qrsh session will suspend the rsh client. EB-2001-12-10-1: Enhancem.: Memory optimization for zombie jobs THE SPOOL FILE FORMAT FOR JOBS (CULL JOB STRUCTURE) HAS BEEN CHANGED JG-2001-12-10-1: Testsuite: fixed qrsh test: - accounting wrongly reported errors - terminate test did no longer work for remote hosts CR-2001-12-10-1: Feature: remove unnecessary libs linked to commd JG-2001-12-10-0: Bugfix: qrsh/qlogin: if client command (rsh/rlogin/telnet) exited abnormaly (signal, coredump etc.), the corresponding job will be deleted, assuring that no tasks keep running on the execution side. CR-2001-12-10-0: Testsuite: using catch to prevent testsuite abort when a test produces a tcl error. EB-2001-12-10-0: Bugfix: Fix in sgeee scheduler where pending tickets are moved into the template for pending tasks Issue: None (Fix of Enhancem.: EB-2001-12-04-0) AH-2001-12-07-0: Cleanup: Improved diagnosis possibilities when commd is blocked NIS/DNS resolver library calls. SD-2001-12-07-1: Bugfix: Fixed ln command in distinst when creating security distribution. Issue: None (Fix of Cleanup: SD-2001-12-06-4) EB-2001-12-07-0: Bugfix: Reanimated code for 'qconf -sep' Issue: None (Bug only appeared in beta1) JG-2001-12-07-0: Bugfix: scheduler could dump core in SGE systems with usersort and maxujobs activated, if no pending jobs existed. Issue: None (Fix of Enhancem. JG-2001-12-06-1) SD-2001-12-06-4: Cleanup: Cleaned up the security gss modules for Kerberos and DCE support. Fixed a bug in get_cred.c. Updated the HTML documentation on using the GSS security modules. Updated aimk, Makefiles and distinst to fully support security modules or security binaries distributions. EB-2001-12-06-3: Bugfix: qstat -s r didn't show running jobs Issue: None (Fix of Enhancem.: EB-2001-12-04-0) JG-2001-12-06-2: Testsuite: Adapt qmaster size test to use newer spawn function open_remote_spawn_process. JG-2001-12-06-1: Enhancem.: Performance improvement in scheduler if the number of jobs/user is limited schedd config, parameter maxujobs. EB-2001-12-06-2: Bugfix: Jobs submittet with start time in future where never started Qalter -hold_jid failed for unenrolled pending jobs Issue: None (Fix of Enhancem.: EB-2001-12-04-0) CR-2001-12-06-1: Enhancem.: commd will only allow kill, trace, dump and unregister commands when client (commdcntl) is using reserved ports. Only a commd which was not started as root will accept connections from other ports. commdcntl can be switched to unsecure mode with -U command line parameter. CR-2001-12-06-0: Testsuite: modifications for new commdcntl -k enhancements. EB-2001-12-06-1: Enhancem.: 'max_aj_tasks' defines the maximum number of allowed tasks within an array job. (75,000) changed default value for 'max_aj_instances' (2000) JG-2001-12-06-0: Bugfix: Cull hash implementation: When cull objects were received and unpacked, all fields that did not have a hash table in the original had a unique hash table created. EB-2001-12-06-0: Bugfix: It was not possible to modify the hold state of tasks without task id specification. (e.g. qalter -h u failed) Issue: None (Fix of Enhancem.: EB-2001-12-04-0) JG-2001-12-05-0: Testsuite: Adapt qrsh test to use newer spawn function open_remote_spawn_process. EB-2001-12-05-0: Bugfix: Master got pending tickets for jobs which where already started. Issue: None (Fix of Enhancem.: EB-2001-12-04-0) SD-2001-12-04-1: Enhancem.: SGEEE - Added support for qconf -clearusage to reset sharetree usage back to zero. JG-2001-12-04-0: Bugfix: shepherd: async commands had no valid filehandles for stdin/stdout/stderr. This could cause the async command, e.g. a terminate method, to fail. EB-2001-12-04-0: Enhancem.: Memory optimization for array jobs AH-2001-12-03-0: Bugfix: Resolve hostname contained in 'act_qmaster' file before comparing it with me.qualified_hostname JG-2001-12-03-0: Testsuite: parsing of qstat -ext output did not return correct cpu value. AA-2001-11-30-4: Feature: Security changes: - common/complex.[ch] -> sge_complex.[ch] due to name conflict when compiling under linux JG-2001-11-30-1: Enhancem.: Event client mechanism: - any user may start an event client from an admin or submit host - changing event client parameters (subscription or event delivery interval) does not take immediate effect, but a number of changes can be made effective by a single qmaster communication. JG-2001-11-30-0: Cleanup: made a number of function parameters const CR-2001-11-30-1: Bugfix: qrsh error with empty job parameters Issue: #110 Bugtraq: #4533448 AA-2001-11-30-4: Feature: Security changes: CSP seclib changes CR-2001-11-30-0: Bugfix: qsub parsing does not handle blanks on job args correctly Issue: #93 Only fixed qmaster output text ("your job X (...) has been submitted."), problem of lost blanks does not appear any more. Maybe someone else fixed it in an other check in. AA-2001-11-30-3: Feature: Security changes: - Makefiles, aimk, aimk.site additions for CSP and Kerberos security AA-2001-11-30-2: Feature: Security changes: - Kerberos Makefiles, make it compilable with const change AA-2001-11-30-1: Feature: Security changes: - qtcsh Makefiles removed EXTRALIBS in gethost buildrule for compilation in sec modes AA-2001-11-30-0: Feature: Security changes: - introduced new product modes sge-csp and sgeee-csp belonging to feature FEATURE_CSP_SECURITY - replaced sec_subst.h and krb_subst.h mechanism by runtime selection of security mode. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! THE send_message[_pb] and receive_message calls have been replaced by gdi_send_message[_pb] and gdi_receive_message. They are responsible for using the correct mechanism. Please use these functions, if the underlying security layer shall be part of the communication. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! As an improvement a communication context struct could be introduced and initialized in the setup phase. To support the sec mechanisms compile with aimk -secure or aimk -kerberos To run in secure mode the product_mode file contains either sge(ee)-csp or sge(ee)-kerberos. SD-2001-11-29-1: Bugfix: SGEEE scheduler - Moved max_pending_functional_jobs short-circuit to outer loop of pending functional jobs algorithm to correct problem where certain pending jobs were not considered for scheduling. (read_write_host.c) removed unused hostname variable JG-2001-11-29-0: Feature: Event client mechanism: - added new events for - admin hosts - submit hosts - configuration changes - calendar - schedd info - manager - operator - all event clients subscribe the events SHUTDOWN and QMASTER_GOES_DOWN and cannot unsubscribe these events scheduler: - use subscription mechanism - do not poll global configuration in each scheduling interval AH-2001-11-29-2: Removed unused exec host spool entries 'number_of_jobs' and 'real_host_name' AH-2001-11-29-1: Bugfix: qmaster logging "enrolled but leave_commd failed ..." Bugtraq: #4532798 Issue: #108 AS-2001-11-29-0: Doc: pseudo output variable $TASK_ID had wrong name $SGE_TASK_ID. AH-2001-11-28-1: Feature: Enhanced accounting(5) file format for max vmem. and PE taskid string. AS-2001-11-28-0: Bugfix: remove unused and obsolete command line parameters in qmaster,scheduler" -noread-argfile -nostart-schedd -nowrite-argfile -truncate-argfile and scheduler: -k -salg Bugtraq: #4532233 Doc: updated man pages sge_qmaster.8, sge_schedd.8 accordingly AH-2001-11-28-0: Feature: added enforce_project switch to cluster configuration removed dead code on QSI/qstd. This requires manual addition of a 'enforce_project' field into the global configuration. CR-2001-11-28-1: Testsuite: Conformation of the testsuite because of changes from CR-2001-11-28-0, Issue: #79, Bugtrag: #4531719 CR-2001-11-28-0: Bugfix: Reject changing of qmaster_spool_dir, execd_spool_dir admin_user, default_domain and ignore_fqdn in running cluster. Issue: #79 Bugtraq #4531719 AS-2001-11-26-0: Bugfix: execd now exits on SIGTERM also when commd is down Bugtraq: #4531720 JG-2001-11-26-3: Feature: Added a subscribe/unsubscribe mechanism to the event client interface. The current implementation does not yet allow dynamic (un)subscription, but only before a (re)registration. Sort output of qconf -secl by event client id. Insert qevent to the Grid Engine program and options handling. CR-2001-11-26-0: Testsuite: Conformation of the testsuite because of changes from AH-2001-11-23-1, Removed unused configuration parameters AH-2001-11-26-1: Bugfix: Fix 'minus_n=0' problem in dist/mpi/rsh wrapper script Bugtraq: #4532155 Build: Fix error in generating rmon lib dependencies AH-2001-11-26-0: Bugfix: Fix qmaster problem with not existing task spool directory when task file is written. Bugtraq: #4532157 JG-2001-11-26-2: Feature: Made qconf -secl use the new lHost datatype and access functions (see CR-2001-11-23-1). AA-2001-11-26-1: Bugfix: Completion for qmon (-> AH-2001-11-23-1) JG-2001-11-26-1: Bugfix: Fixed a (non critical) read to uninitialized memory in the code retrieving memory load values on Solaris. JG-2001-11-26-0: Feature: Added qconf options -secl (show event client list) and -kec (kill event client). AH-2001-11-23-1: Bugfix: Removed unused configuration parameters in queue configuration and scheduler configuration Issue: #71 Bugtraq: #4431000 This requires manual removal of certain parameters before restarting daemons - max_migr_time - migr_load_thresholds - max_no_migr - klog - reauth_time from all queue configurations stored in $SGE_ROOT/default/spool/qmaster/queues/* and - maxgjobs from the scheduler configuration in $SGE_ROOT/default/common/sched_configuration This change is not yet done in qmon but only in the core system. Please don't try to compile qmon until qmon has been updated accordingly (aimk -no-qmon). AH-2001-11-23-0: Feature: Enhanced queue priority range to -20 to +20 changed data type of queue priority from ulong to string removed priority attribute from queue complex CR-2001-11-23-1: Feature: Implementation of the lHostT cull data type. Modified each cull list where a hostname is stored. The lHostT data type is implemented like the lStringT type, but only used for hostnames. IssuZilla: #63 Bugtraq: #4531708 CR-2001-11-23-0: Testsuite: changed timeout behaviour in procedure open_remote_spawn_process CR-2001-11-22-0: Testsuite: rework on auto_reschedule test faster handle_vi_edit procedure AH-2001-11-21-1: Feature: Added NO_REPRIORIZATION setting in qmaster_params to allow for diabling the PTF module. Without this setting the 'priority' parameter in queue_conf(5) has no effect in a SGEEE system. JG-2001-11-20-0: Feature: Generalized event client interface. The former event client interface was very scheduler specific. Only one event client - the scheduler - could register. Now any number of event client can register at qmaster and receive events. A sample event client can be found in clients/qevent. Further cleanup steps and documentation still has to be done. A subscribe/unsubscribe mechanism does not yet exist. AH-2001-11-20-0: Bugfix: Use kstat interface instead of kvm interface for retrieval of cpu percentage load value to fix problems with Solaris 7,8 running 32-bit kernels IssuZilla: #78 Bugtraq: #4528905 SD-2001-11-19-0: Feature: Added support for queue starter_method Issue: #6 AA-2001-11-19-0: Bugfix: Qmon core dump in sharetree for empty sharetree when using Find JG-2001-11-15-0: Feature: Performance/Scalability optimization: Defined hash tables on a number of objects to speed up search operations. Changed the order in which non unique hashed objects are stored - some code relies on getting them in the same order they were stored. EB-2001-11-13-0: Bugfix: Qmaster did not change the reschedule unknown value immediately after using the qconf command. Issue: #92 Bugtraq: #4525963 AA-2001-11-09-0: Build: rename of commd constants to resolve namespace conflicts for perlgui added aimk -perlgui switch to build perlgui security changes work in progress CR-2001-11-08-0: Testsuite: - adapted default haftime scheduler parameter from 24 to 168 - new debug parameters chanched behaviour of performance test when calling open_remote_spawn_process - fixed - changed deviation value for share tree check - from 1% to 2% CR-2001-11-08-0: Testsuite: - new debuging options: menu -> new points 20 and 21 (shutdown/restart scheduler and master) command line -> master_debug, schedd_debug and display master and scheduler can now be started in debug mode - new procedure: wait_for_unknown_load - (sge_procedures.tcl) JG-2001-11-08-1: Feature: New qconf options: -Auser: add user from file -Muser: modify user from file Corrected sorting of options in help output. Issue: #89 JG-2001-11-08-0: Bugfix: Updated recent GDI versions for version check and error handling in qmaster. ----------------------------------------------------------- AS-2001-11-07-0: Created V53_beta1_TAG and V53_beta1_BRANCH Released SGE 5.3beta1 JG-2001-11-07-2: aimk: recognize the necessity to set -DHAS_IN_PORT_T for the LINUX6 platform dependent on the libc version. AA-2001-11-07-0: assure that cull_hash_free_descr get a valid descr JG-2001-11-07-1: Bugfix: Retrieving a string value from a cull object field and setting this value again for the same field produced a memory access error. JG-2001-11-07-0: Adjust MAKEOPT for aimk option -qmake to reflect latest changes to qmake. EB-2001-11-06-1: Fix: qdel of pending job task array took a long time CR-2001-11-06-0: Testsuite: - comparing reset_scheduler setup values on startup with the default from installation - additional testing procedures for sharetree test - setting default value for rlogin_max_open_connections down to 20 - procedure reset_scheduler setup value update EB-2001-11-06-0: classic_sgeee_scheduling=1 => share_override_tickets=1 => share_functional_shares=0 => share_deadline_tickets=1 AA-2001-11-06-0: GE version of intro,logo.xpm, colors reduced AA-2001-11-05-3: GE version of intro,logo.xpm AA-2001-11-05-2: Fix for core dump if no submit/admin host in ticket dialogue AA-2001-11-05-1: copyright for GE AA-2001-11-05-0: obsolete pixmaps removed AA-2001-10-31-0: copyright notice EB-2001-11-05-0: Bugfix: Ckpt jobs and jobs exiting with exit state 99 could meet an old instance of the same job if the scheduler was to fast. AH-2001-11-05-0: Fixes with qstat -pe and qselect -pe - 'all' keyword in PE's queue_list was not honoured - non-parallel queues were selected Bugtraq: #4584673 JG-2001-11-02-1: qstat -j aborted when writing usage of pe tasks JG-2001-11-02-0: qmake: Does not need to export variable PWD when executing remote tasks (qrsh option -v PWD) - PWD is setup correctly by execd/shepherd. CR-2001-10-31-0: Testsuite: Added new test for share tree in checktree/system_tests/grd/sharetree/check.exp JG-2001-10-31-3: Bugfix: Parsing of command line switches -ac, -dc, -sc and -v failed if the value of the passed variable contained spaces, e.g. -v EDITOR="gvim -f",DISPLAY=elendil:0.0 created the environment variables EDITOR=gvim, -f="" and DISPLAY=elendil:0.0 JG-2001-10-31-2: -cwd option for tightly integrated pe tasks (qrsh -inherit) did not overwrite the jobs -cwd settings. JG-2001-10-31-1: Bugfix: qmon dumped core when displaying empty job args JG-2001-10-31-0: Testsuite: Bugfix in submit_hosts test, did not work, if master host and submit_hosts test host had different architecture. EB-2001-10-30-1: Disabled modifications done in EB-2001-10-30-0 JG-2001-10-30-2: Testsuite: Accept new versions - SGE 5.3beta1 - SGEEE 5.3beta1 JG-2001-10-30-1: Bugfix: Could not start jobs with empty commandline arguments AA-2001-10-30-1: Fix for functional ticket dialogue Success message AA-2001-10-30-0: Fix for defaultdepartment callback sensitivity of clusterparams IZ 81 JG-2001-10-30-0: qtcsh could not be compiled on newer linux systems (SuSE 7.2) due to changes in the operating system include files EB-2001-10-30-0: Modifications for sgeee scheduler - Changed initial ticket count for started jobs - Normalized pending job tickets - Changed several parameter for sgeee scheduler JG-2001-10-29-0: Removed explicit hash tables on job lists in qmaster and scheduler - they are no longer needed, as cull hashing functionality was enabled for job lists. AH-2001-10-29-0: Qmaster: - cancel resend of jobs to execd on forced job deletion AA-2001-10-29-1: security files that are no longer needed removed AA-2001-10-29-0: Fix for weired characters in helper dialogs (PE,Project Configi, Calendar, Cluster Config, Complex Config) Queue Control Reschedule functionality added (corresponds qmod -r ) Cluster config Reschedule Unknown bug fixed Job Custom dialogue core dump fix EB-2001-10-29-0: Fix for sgeee scheduler - empty sharetree && sharetreetickets > 0 caused fatal exit of scheduler CR-2001-10-26-1: Testsuite: - bugfix for path_alias check: problems with file creating on NFS directory. The check will now make sure that the correct/updated path_alias files are used (diffing file output on each execution host in setup procedure). CR-2001-10-26-0: Source: - Added variable initialization - Removed double defined message Testsuite: - using cvs -P parameter for source update - issue #8 (one rlogin at startup): - new parameter "close_rlogin" for compatibility EB-2001-10-25-0: Modified qdel, qmon, qmaster - removed job attribute vom "Job Customize" dialog - bugfix: qdel -u - moved messages into msg_*.h file JG-2001-10-25-0: Fixed some memory leaks JG-2001-10-23-0: Fixed a bug in the implementation of non unique hash tables: The key data has to be duplicated in the hash table. Therefore a duplication function has to be passed when creating a new hash table. JG-2001-10-22-0: Fixed a memory access error in the cull hash implementation EB-2001-10-20-0: Added 'const' in libraries JG-2001-10-18-0: Execd: - bugfix: reporting of a parallel array jobs pe task in tight integration could fail. - use cull hashing iterators for searches in job list JG-2001-10-17-0: (Issue #57) Sparc Linux port (32 bit). Grid Engine platform slinux. The port is restricted to SGE as the pdc module had to be disabled for slinux (aimk variable COMPILE_DC). JG-2001-10-16-0: Implemented the use of hash tables for cull lists. The list definition (libs/gdi/sge_*L.h) contains information, for which data fields hash tables shall be generated. Hash tables are created and maintained by the cull list access functions (lCreateList, lAppendElem etc.) and are used by the cull search functions (lGetElem*). Defined default hash tables for the most used cull lists. Increased the GDI version - the spooling file format and the message format changed. EB-2001-10-11-1: Changed MAX_SEQNUM from 999,999 to 9,999,999. - changed qstat output format EB-2001-10-11-0: Changed spool behaviour for (array) jobs - beared down the 32765 subdirectory limit - added some tuning parameters for job spooling - collected duplicated code used in master and execd CR-2001-10-10-1: Testsuite: - closed issue # 9 (Create a new test for file parsing) CR-2001-10-10-0: Testsuite: - new command line option: mfile FILE (don't mail messages, write it to file) - new optional parameter for procedure "wait_for_jobstart" (do a qconf -tsm before waiting for jobstart) CR-2001-10-09-2: (sge_rusage.c) - Issue #68 (The accounting file handles only entries < 2048 chars/line) CR-2001-10-09-1: (rmon_macros.c) - don't use vnsprintf() for ALPHA4 and IRIX6 CR-2001-10-09-0: (parse_job_cull.c) - Issue #67 (long lines in sge_request files causes bus error in qsub) - removed some local char str[1024] definitions and replaced it with error_string buffer one function level above (rmon_macros.c) - using vnsprintf() system call as default (instead of vsprintf) This should prevent a buffer overflow error JG-2001-10-09-0: Added statistics for our hashtable implementation that gives following info: - size of hashtable and number of elements stored - number of empty hash chains - maximum and average chain length When a hashtable is resized and log_level is set to log_debug, statistics and profiling information is output to messages files. JG-2001-10-08-0: Enabled options -h and -hold_jid for qrsh Bugtraq #4511859 AH-2001-10-05-0: Removed warning message in execd EB-2001-10-05-0: minor modifications - remove/displace code - comments/formatting JG-2001-10-04-0: Generalized our hash table implementation to support any data type for keys. Hash table uses hash and compare functions to be specified when a table is created. Hash and compare functions for strings and u_long32 are provided. EB-2001-10-04-0: (Issue #7) - changed spool behaviour for (array) jobs AH-2001-10-02-0: Schedd: Moved code handling 'forced' attributes to a section handling other static matchings. Static matchings can be done independently of consumables/load values and thus needs be done less frequent SD-2001-10-01-0: Updated the SGEEE pending job algorithm. WARNING - WARNING - WARNING - WARNING - WARNING - WARNING This update changes the binary job structure spooled to disk. This means that if you are already running with the open source code, you will need to delete all pending and active jobs before running with the new SGE binaries. This is true if you are running either SGE or SGEEE. CR-2001-09-26-0: Testsuite: - new setup file version (new parameter "list_of_submit_only_hosts") - bugfixes in path_alias test - bugfix/enhancement in set_config procedure - Issue #12: New test: submit_hosts test AH-2001-09-21-0: Improved array-job performance in schedd CR-2001-09-21-0: Testsuite: - New procedure in file_procedures module: create_path_aliasing_file() - Issue #54 : New test: sge_aliases file CR-2001-09-18-0: Testsuite: - sending CR (carriage return) before starting remote command (to stop eventually switched on correct questions) on rlogin call - wait longer before sending password (some login daemons don't accept fast password typing) CR-2001-09-17-0: Fixed issue #60 "COMMPROC ALLREADY REGISTERED" error AH-2001-09-14-2: Fixed error with job array causing dispatch of the last array-job being postponed to the follow-up scheduling interval JG-2001-09-12-0: (Issue #58) - scheduler optimization: sorting of hosts is only done when really necessary. - schedd can output profiling information - set schedd_params to profile - some smaller bugfixes AA-2001-09-06-0: Fixed y out of range of allowed Xserver values in QControl This limits the number of displayable qs to 2758 JG-2001-09-03-0: (Issue #51) - PWD was set incorrectly for jobs submitted with qsub -cwd if a path mapping was active (sge_aliases file) - qrsh does no longer use the PWD variable to determine target job working directory but jobs configuration CR-2001-08-31-0: Testsuite: - issue #10 shutdown_system_daemon does not work for some platforms - start_remote_prog procedure can now export user environment on remote host EB-2001-08-30-0: Removed help output from qconf concerning 'meattr' and 'Meattr' (Issue# 50) AA-2001-08-28-0: Fixed task array job handling in qmon submit dialogue (Issue# 48) EB-2001-08-27-0: Fixed segmentation fault in qstat, qhost, qmon (Issue# 48) - large, fragmented job arrays caused segmentation fault in qstat when the task id range list should be printed - hold of a list of array tasks in the job control dialog caused only the last one to be set into the hold-state CR-2001-08-23-2: Testsuite: - using full path name for files to check in procedure check_executable_files CR-2001-08-23-1: Testsuite: - removed add_proc_error call in compile_source procedure (output buffer overflow error). Now the error is only reported in the normal testsuite output. In case of parallel compiling more architectures this error message sometimes appears. CR-2001-08-23-0: Testsuite: - using cvs tag information for compile logfile name JG-2001-08-22-0: (Issue# 43), host_aliases file was not implicitly used from standard location, option -a worked. AA-2001-08-22-0: - bugfix IZ 42, processing of projects/xprojects in SGEEE mode AA-2001-08-17-0: - enhanced scripts/snapshot.csh to use TAG as part of filename JG-2001-08-13-0: fixed/improved handling of aimk option -opt (optimization) at Ron Chen's suggestion. AA-2001-08-09-0: - IZ 33, fixed duplicate free for scheduler conf dialogue code AA-2001-08-07-0: - IZ 32, feature_is_enabled() used incorrectly AH-2001-08-03-0: Accept -N job_name references in -hold_jid list JG-2001-08-02-0: Added external loadsensor script for platform aix43 (util/resources/loadsensors/ibm-loadsensor) JG-2001-08-01-0: Testsuite: Test "cluster performance" failed on hosts reporting a fully qualified hostname. JG-2001-07-31-1: qrsh_starter: Handle (warn and ignore) empty environment variable QRSH_WRAPPER JG-2001-07-31-0: (Issue# 28): performance tests "cluster" and "scheduler" failed in sgeee systems AA-2001-07-27-0: - Merge from branch: core dump fix for qmon_pe start/stop_proc_args JG-2001-07-27-0: (Issue# 22): qdel to non existent job output inaccurate error message JG-2001-07-26-1: Maintrunc version: - set Version string to "5.3 maintrunc" - increased GDI version AH-2001-07-26-0: - improved handling of waitpid() result in execd to trace into the problem of shepherds considered to be finished by failure (Origin 2000, Irix 6.5.10f) - cleaned up on job logging code - added job logging also to Schedd - fixed typo in setfileperm.sh CR-2001-07-26-1: Testsuite: - create protocols directory on compile_source procedure if it is not existing CR-2001-07-26-0: Testsuite: - added compile error output to mail on compile error - added compile output log file generation in protocol dir JG-2001-07-26-0: (Issue# 15): startup_template: replaced fixed path by reference to SGE_ROOT CR-2001-07-25-2: Testsuite: - modified compile mail output CR-2001-07-25-1: Testsuite: - CHECK_EXPECT_MATCH_MAX_BUFFER reset to 640 KB - full_buffer caused not the error in compile_source procedure (removed add_proc_error call in compile_source on full_buffer) (full_buffer themes not to work correctly for multiple spawn/ expect calls) EB-2001-07-25-1: (Issue# 4) fixed build error (LINUX6, ALINUX) EB-2001-07-25-0: (Issue# 5) qmaster - qconf: - qconf -so|sm does not show comments anymore. - Comments will only be inserted into spoolfiles. CR-2001-07-25-0: Testsuite: - reading CHECK_GROUP and CHECK_USER global variables from check.exp file attributes if environment variables GROUP and USER aren't set. CR-2001-07-24-1: uninitialzed variables error/warnings on linux in sge_job.c CR-2001-07-24-0: Testsuite: - CHECK_EXPECT_MATCH_MAX_BUFFER set to 6,4 MB - enable empty lines when reading in configuration files - using env(HOST) only when HOST environment is set JG-2001-07-24-1: qmaster - qdel: - if a job is already in state deleted additional qdel's do not generate additional events to execd - improved error output concerning tasks to be deleted (former output contained only job id), additional messages: job "1595" task "4" does not exist job "1595" task id range "1-4:2" comprises no tasks - fixed bug in deletion of events to be resent JG-2001-07-24-0: Adoc: Function identifiers beginning with "-" (e.g. --Introduction) are not included in the function index CR-2001-07-23-0: Bugfix: qstat -j (Issue#: 2) causes "operation not implemented for target" error EB-2001-07-23-0: Changes some html files - link to contributor agreement - added $ARCH variable to path (gethostname) - deleted empty line in client output - changed text width CR-2001-07-20-0: Testsuite: - qmod reschedule test is not completely implemented - qrsh test showed error because of killing a job script (exist status of expect child not 0) - disabled error message for this open_spawn_process call - shadowd migrate test can will stop if only one execution host is in the cluster AH-2001-07-20-0: - increased GRM_GDI_VERSION for GE 5.3 - fix qmaster core dump caused by old execd's sending load reports w/o REP_seqno