[GE users] prolog leaves queues in Error state

Rene Salmon rsalmon at tulane.edu
Thu Jul 29 14:57:29 BST 2004


Hi list,


I am trying to setup a prolog script to check for number of cpus requested
in an mpi parallel environment and depending on the number either
run the job or exit.  I found this on the list archives.

#!/bin/sh
#**************************

if [ "$NSLOTS" -gt 8 ] ; then
        echo
        echo "*** Maximum number of CPUs exceeded. ***"
        echo
        . /opt/gridengine/default/common/settings.sh
        qdel $JOB_ID
fi



The problem is that every time I submit a job and this prolog script gets
executed the job imediately goes back to the queued state and it leaves
whatever queue hosts it tried to run on in the error state.

I found and issue on this #391.  But it states this is fixed for my
version of SGE.  I am running SGEEE 5.3p5 with

shell_start_mode          unix_behavior

This is my first attempt at setting up a prolog script and restricting
number of CPUs per user so if anyone has a better way please share.

Thanks
Rene




---------------------------------------------------------------------
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