Thread: Job ID Created Twice

Originally posted by: JPummill, Wed Jan 10, 2007 1:56 pm

When submitting a job to the queue ( bsub < lsfscript ), it enters the queue a first time with a given Job ID 100, then it sends me a note saying that the job has been re-submitted and it gets a new Job ID 101.

Is this normal? The problem is that I am "hopping" it over other jobs in the queue with btop, so I end up having to do this twice to get the job to run.

This is my lsfsubmit script...

[jpummil@prospero ~]$ more lsfsubmit
################################################## #
# MPICH-P4 Example Script File #
# Use the /opt/mpich/gnu paths #
################################################## #
#!/bin/sh

#Number of cpus
NCPUS=8

#Output file, will get written in CWD, %J will be translated into the
#actual jobid
OUTFILE=%J.out

export P4_RSHCOMMAND=/usr/bin/ssh

#Run a job
bsub -o $OUTFILE -n $NCPUS -a mpichp4 mpirun.lsf /home/jpummil/xhpl

#######################################
# End of MPICH-P4 Example Script File #
#######################################

Originally posted by: ddunlap, Thu Jan 11, 2007 4:42 pm

This is not normal. The reason for this is that you have another "bsub" embedded in lsfscript. In other words, you are doing this:

bsub < lsfscript

That creates job #100. But, all that job does is this:

bsub <args> mpirun.lsf ...

That one is job #101.

You should just run:

./lsfscript

My $.02,
Dale
_________________
Dale Dunlap
Technical Consultant

Originally posted by: JPummill, Thu Jan 11, 2007 9:52 pm

Thanks Dale!

It is still pending even after a btop as all of the slots are currently full, but what you said made perfect sense after I examined it. Time will now tell, huh?

-Jeff