[gridengine users] Job priority reported by 'qacct -j JOBID'

Reuti reuti at staff.uni-marburg.de
Thu Nov 22 16:44:26 UTC 2012


Am 22.11.2012 um 17:26 schrieb Gowtham:

> Following this article:
> 
>  http://www.gridengine.info/2008/03/13/clever-job-prioritization-tip/
> 
> I have edited '/opt/gridengine/default/common/sge_request'
> to include the following line at the end
> 
>  -p -100
> 
> 
> When I set a job to have relative priority of say, -30,
> in the submit script to the queue, it seems to start
> running sooner than a job that didn't have an explictly

Default should be 0.

But it depends on the settings in the scheduler (`qconf -msconf`), how this setting will influence the order of jobs. There is a man page `man sge_priority` with the formular for the computation, which you can check with:

$ qstat -urg
$ qstat -pri
$ qstat -ext


> set relative priority OR a job that had a value of -50.
> 
> Also, when the job is running, output of 'qstat -j JOBID'
> reports the this relative priority value correctly.
> 
> So, I do know that this thing is working.
> 
> 
> However, after the job has long been completed, the output
> of 'qacct -j JOBID' seems to show priority as 0 for all
> jobs.

This is the "priority" inside the queue definition - i.e. the nice value which was used for the job.

-- Reuti


> Is this a feature/bug?
> 
> 
> Best,
> g
> 
> PS: I notice this on our Rocks 6.0 and Rocks 5.4.2 clusters
>    both using SGE 6.2u5
> 
> --
> Gowtham
> Information Technology Services
> Michigan Technological University
> 
> (906) 487/3593
> http://www.it.mtu.edu/
> 
> _______________________________________________
> users mailing list
> users at gridengine.org
> https://gridengine.org/mailman/listinfo/users





More information about the users mailing list