This document describes the automatic user concept introduced in
Grid Engine 6.0.

The Grid Engine Enterprise Edition scheduling
policies allow specifying priorities based on users. Previous
versions of Grid Engine Enterprise Edition required the manual
creation of user objects in order to have a place to specify the
relative importance of users and to store computer usage (CPU,
memory, and I/O) for users. Grid Engine 6.0 supports the
capability to automatically create user objects based on
predefined default values. This can significantly reduce the
administrative burden for sites with a large number of users.
User objects will be created automatically if the enforce_user
cluster configuration parameter is set to "auto".  The
auto_user_oticket, auto_user_fshare, and auto_user_default_project
cluster configuration parameters will be used to set the respective
user object attributes.  The cluster configuration parameter
auto_user_delete_time will control when automatically created
user objects will automatically be deleted.  If the
auto_user_delete_time is set to zero, the user objects are
permanent. Otherwise, user objects will be deleted after the
user has had no active jobs for the specified time in
auto_user_delete_time.  The deletion time for each automatic
user is maintained in the the user object and can be examined
using qconf(1) or qmon(1). The deletion time will be zero for
permanent users.  If an automatically created user has active
or pending jobs, the deletion time will be one.  If the
automatically created user does not have active jobs, the deletion
time will be set to the time (in number of seconds since January 1,
1970) when the user will be deleted.

In conjunction with this change, Grid Engine 6.0 also supports
non-project based default user nodes in the share tree. When the
share tree policy is being used, a job's priority is determined
based on the node the job maps to in the share tree. Users
which are not explicitly named in the share tree will map to the
default node, if it exists.  Previously, default nodes could only
be specified in a project branch. Allowing the specification of
non-project default nodes allows a simple share tree to be created
with a single default node to allow user-based fair sharing. This
provides a capability which is similar to the user_sort capability
in Grid Engine versions prior to 6.0. The qmon(1) utility has also
been updated to display "virtual" nodes for all users which
map to the default node.  This allows users or administrators
to examine the usage and fair share scheduling parameters
for users which map to the default node.