[GE users] specifying max_errors for qsub

Petla, Raghuram_Murthy Raghuram.Murthy.Petla at deshaw.com
Wed Mar 9 15:03:29 GMT 2005


Hi Ron,

here fail means, if job fails during execution.

More detailed explanation:

Assume that I have submitted a job with 200 jobs and I want to set 10 as
max_errors_allowed.
Now SGE schedules each job when a node is free. Assume that among 50
jobs run 10 jobs exited with non-zero values.
I consider those 10 as failed commands. If one more job exits with
non-zero value, then total failed commands are more than 10, so I want
to delete 52:200 jobs from being scheduled.

I just want to check is there any option for sqsub to do this.

Currently whay I am doing is, I am submitting the job and constantly
running qacct command to get the exit status of each job and using qdel
if number of failed commands are more than maximum allowed.

I want to avoid constant calling of qacct, as it causes CPU hog.

Thanks,
-Raghuram


> Each qsub request should be independent of each other.
>
> And by "fail", you mean it fails to submit or the job
> itself fails during execution?
>
> -Ron

--- "Petla, Raghuram_Murthy"
<Raghuram.Murthy.Petla at deshaw.com> wrote:
> Hello,
> 
> I am submitting around 200 jobs using qsub.
> Considering that a job fails
> if it exits with noz-zero, I want to specify
> max_allowed_errors, so that
> qsub should abort remaining jobs if there are more
> number of fail jobs
> than maximum allowed.
> 
> How can I achieve this?
> 
> Thanks
> -Raghuram
> 
> 
>
---------------------------------------------------------------------
> To unsubscribe, e-mail:
> users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail:
> users-help at gridengine.sunsource.net
> 
> 


	
		
__________________________________ 
Celebrate Yahoo!'s 10th Birthday! 
Yahoo! Netrospective: 100 Moments of the Web 
http://birthday.yahoo.com/netrospective/

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list