[torquedev] Asynchronous jobs through qrun not qsub

Al Taufer ataufer at clusterresources.com
Tue Nov 20 09:17:23 MST 2007


In my original post I incorrectly stated that it was qsub that we wanted 
to change, in reality it is qrun that we change.  We would use 'qrun -a' 
(that option doesn't yet exist in qrun).

Some schedulers such as Moab already use the Torque asynchronous job run 
capability which uses the pbs_asyrunjob() interface, but for some reason 
it was never added to the Torque commands.  The asynchronous run mode 
causes the pbs_server to reply with an ack before it sends the job to 
the mom.  In theory it should allow higher throughput when launching jobs.

Sorry about the confusion,
Al
> We would like to add the ability to submit an asynchronous job through 
> qsub.  We would add a '-a' parameter which signifies that the job 
> would be asynchronous.  Default behavior for qsub would remain the 
> same as it currently is.
>
> Any comments or concerns?
>
> Thank you,
> Al
>
> _______________________________________________
> torquedev mailing list
> torquedev at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torquedev


More information about the torquedev mailing list