Bill,
Here is an earlier email conversation between you and Dave Woods that describes an issue which is potentially related to your current one.
Here, some odd characters in the script may have caused it to fail.
In general, once you submit a script it should go from the queuing mode "Q" to the run mode "R" pretty much immediately, since the cluster is not too busy right now (check with qstat). When the job has been accepted to "run" there will be files written upon its termination, regardless of whether the execution of the program(!) failed or succeeded.
If you submit scripts, similar to the one below, to the queue repeatedly you should be getting error messages for each of the submissions, in case the program fails to execute (in the directory you submitted the job from with qsub).
So, to reproduce your observation that multiple jobs can't be launched concurrently, I suggest to submit a number of jobs through a script similar to the one below and monitor the queue via qstat. Upon acceptance into the run status, if one or more fail to execute, I will need to examine the output .o and .e files that have been created to make an assessment of what has gone wrong.
Please let me know if you need more details.
Jens
-----Original Message-----
From: Research Computing Support [mailto:[log in to unmask]] On Behalf Of Woods, David M. Dr.
Sent: Monday, June 23, 2008 7:38 PM
To: [log in to unmask]
Subject: Re: script for cluster
Bill,
I think I see what the problem is, but I'm not sure what caused it. Our system administrator, Robin, may have some ideas on the cause.
I copied your file to my area and when I look at it with an editor, I see that the last line is actually
stata-se| -b| do| fama_frenchreg.do|
I think the "|" symbol is indicating an odd character because when I looked at the file using other tools, it looked OK.
I'm wondering how you created the file - that might help us understand where these odd characters got introduced.
I can ask Robin to fix it, or you could edit the file on the cluster. If you want to do this, go to the directory where the file is located and issue the command:
nano batchdo.job
Use the arrow keys to move to the last line and use the delete key to delete all of the text and then re-enter it. To save the file use the Control-o (control and letter o key) and then do Control-x to exit.
Dave
________________________________________
From: Research Computing Support [[log in to unmask]] On Behalf Of Even, Bill [[log in to unmask]]
Sent: Monday, June 23, 2008 7:01 PM
To: [log in to unmask]
Subject: script for cluster
I'm submitting the following script on the cluster and failing to get the job to execute. The only output created says
/usr/local/torque/2.1.6/var/spool/torque/mom_priv/jobs/184357.muln.SC: line 9: stata-mp -s do fama_frenchreg.do : command not found
Here's the script that I ran.
#!/bin/bash -l
#PBS -N ffreg
#PBS -l nodes=1:bigmem:ppn=2
#PBS -l walltime=10:0:0
#PBS -j oe
#PBS -q stata
module load stata
cd /shared/evengroup/projects/partdir/form5500/analysis
stata-se -b do fama_frenchreg.do
Thanks!
Bill Even
|