[GE users] Job rescheduled but not killed

kain2log gilbert at rldp.com.ph
Mon May 24 02:19:13 BST 2010


> Am 21.05.2010 um 03:03 schrieb kain2log:
> 
> >> Hi,
> >> 
> >> Am 20.05.2010 um 11:34 schrieb kain2log:
> >> 
> >>> I want to reschedule a running job, 
> >>> then that job must be killed in order for the slot to get a new job.
> >>> 
> >>> first I enable the job to restart, then reschedule;
> >>> 
> >>> qalter -r y $JOB_ID
> >>> qmod   -rj  $JOB_ID
> >>> 
> >>> The problem is that when I do this, sometimes job is killed and sometimes it is not.
> >>> 
> >>> The job gets an Rq status, but when I check the processes, sometimes the job is still running or active.
> >>> 
> >>> How would I be sure that a rescheduled job was really killed?
> >>> Did I miss a command?
> >> 
> >> there was an issue which was solved in the last update. Which version are you running?
> >> 
> >> -- Reuti
> > 
> > We are running Grid Engine 6.2u5.
> 
> I thought of this issue:
> 
> http://gridengine.sunsource.net/issues/show_bug.cgi?id=1521
> 
> Hence it should be fixed. You also observe, that you have the same job then twice in the cluster?
> 
> -- Reuti

Yes, that issue exactly what happens in my case.
The strange thing is that sometimes "qmod -rj <job_id>",
really kills and reschedule the task.

I have not noticed that the same job run twice in the same cluster, though it maybe possible.

When I reschedule a job, I manually double checks if it was killed or not, 
A new job is executed, and this new job will eventually be an error if the reschedule job will not be killed because the license is still being used by the previous(rescheduled) job.

Is there a way to ensure that a rescheduled job was really killed?

-- Kain
> 
> 
> > --Kain
> > 
> >> 
> >> 
> >>> Thanks,
> >>> 
> >>> Kain
> >>> 
> >>> ------------------------------------------------------
> >>> http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=257959
> >>> 
> >>> To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].
> > 
> > ------------------------------------------------------
> > http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=258036
> > 
> > To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=258318

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list