Re: [Condor-users] how to resrict job run time

[Dan Bradley <dan@xxxxxxxxxxxx>]

I recommend the following expression for removing jobs that run for too 
long:

maxRunTime = 60
periodic_remove = JobStatus == 2 && \
 ( (CurrentTime - EnteredCurrentStatus) + \
   RemoteWallClockTime - CumulativeSuspensionTime < $(maxRunTime) )

Warning: that may be useful, but it is not quite "correct" for several 
reasons:

1. It does not take into account suspension time in the current run 
attempt, only previous run attempts.

2. It does not take into account whether previous run attempts were 
preempted without saving a checkpoint.

Given the attributes that are published into the job ClassAd, I see no 
way to correct those cases in current versions of Condor.

--Dan

Jens Harting wrote:

> On Wed, 25 Oct 2006, Dan Bradley wrote:
>
>  
>
> > Use periodic_remove or periodic_hold in the job submission file.  See 
> > the condor_submit manual for details.
> >
> > --Dan
> >    
>
> I did try that, but found a problem:
>
> Adding the following lines to my submission script do work:
>
> maxRunTime               = 60
> periodic_remove = (RemoteWallClockTime > $(maxRunTime))
>
> However, the value for RemoteWallClockTime will only be updated if the job 
> is being suspended. Since it sometimes happens that a job runs on a 
> machine without being preempted much longer than $(maxRunTime), the 
> expression for periodic_remove cannot evaluate to true.
>
> Is there any way to tell Condor to update RemoteWallClockTime more 
> frequently?
>
> jens
> _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>
> The archives can be found at either
> https://lists.cs.wisc.edu/archive/condor-users/
> http://www.opencondor.org/spaces/viewmailarchive.action?key=CONDOR
>