Skip to Main Content

The University of Tennessee

Newton header banner

Frequently Used Tools:




Job priority

Once the job is submitted, it is placed in a "pending" state until the Grid Engine can find an available queue in which to execute it. When a resource becomes available, the Grid Engine sorts the list of pending jobs according to their priority values and attempts to execute the highest priority job. A job's priority is based on a number of factors:

  • Your group's historical resource usage
  • (optionally) Your own historical resource usage

The Grid Engine monitors the historical resource usage (number of CPUs used) of every group and user on the system. It uses a 7 day half-life to weight the resource use and compares the resources used to the amount of resources that the group has been guaranteed according to the membership agreement. This means that a group which has used less than its resource allocation will have a relatively higher priority than a group that has been using more than the resource allocation. You can view your group resource allocation (interpreted as number of CPUs) as follows:

[myusername@newton ~]$ id
uid=5114(myusername) gid=5104(mygroup) groups=5104(mygroup)
[myusername@newton ~]$ qconf -sstnode mygroup
/mygroup=128

Resource usage can also affect your job on the individual user level, but only if you group has elected to use this feature (please ask an administrator if you would like this enabled). This may be useful if you wish to assign a proportion of the group's allocation to specific group members. Otherwise, all group jobs will have the same priority and will be executed in the order that the jobs were submitted.

Back to Using The Grid Engine