WebMO - Computational chemistry on the WWW
Recent news

WebMO 18.1 is now available for free download!

WebMO 18.1 Pro and Enterprise have a variety of additional features and is available for purchase.

The WebMO app is now available for iOS and Android.

Signup for WebMO workshops this summer 2018

January 25, 2020

External batch submission failing (SL... Log Out | Topics | Search
Moderators | Edit Profile

WebMO Support Forum » Running WebMO Jobs » Job Manager » External batch submission failing (SLURM) « Previous Next »

Author Message
Liam Forbes
Unregistered guest
Posted on Wednesday, August 01, 2018 - 8:23 pm:   

I am attempting to submit jobs via SLURM from a VM running WebMO to an HPC cluster running SLURM. I seem to be running into a problem with the qsub command, and can't figure out the problem based on the run_log.

Initially I was getting the following error:

Value "t1small" invalid for option p (number expected)
Unknown option: nodes
Unknown option: tasks-per-node
qsub [-a start_time] [-A account] [-b y|n] [-e err_path] [-I] [-l
resource_list] [-m mail_options] [-M user_list] [-N job_name] [-o
out_path] [-p priority] [-pe shm task_cnt] [-P wckey] [-q destination]
[-r y|n] [-v variable_list] [-V] [-wd workdir] [-W
additional_attributes] [-h] [script]

I was able to resolve the first three messages by changing "-p '$queue'" to "-q '$queue'" in public_html/cgi-bin/webmo/daemon_pbs.cgi. I also changed "--nodes=$nodes --tasks-per-node=$ppn" to "-l nodes=$nodes:ppn=$ppn" in the same line. However, I'm not sure what's generating the usage message, and the job fails immediately.

If I cd into the job directory, I'm able to execute the qsub from the command line based on the comments at the top of pbs_script.sh.

webmo % pwd
webmo % /opt/scyld/slurm/bin/qsub -o /home/webmo/webmo/loforbes/3/pbs_stdout.test -e /home/webmo/webmo/loforbes/3/pbs_stderr.test -q 't1small' -l nodes=1:ppn=1 pbs_script.sh
webmo % qstat
Job id Name Username Time Use S Queue
------------------- ---------------- --------------- -------- - ---------------
245272 ARCTIC4 nobody 03:06:26 R t2standard
245795 MOM6-CCS1 nobody 00:05:37 R t2standard
245805 MOM6-CCS1 nobody 00:03:49 R t2standard
245807 myscript.sbatch nobody 00:03:33 R t2standard
245827 pbs_script.sh webmo 00:00:00 R t1small
webmo % cat run_log
Executing script: ./run_gaussian.cgi
Creating working directory: /center1/COMPCHEM/webmo/webmo-1945/3
Script execution node: n7
Job execution node(s):
Executing command: /usr/local/pkg/gaussian/gaussian-09.D.01/g09/g09

I can see the Gaussian app running on the cluster compute node and eventually the job finishes properly, although it doesn't show up in WebMO that way. I'm not sure what to try next to diagnose the usage message.


-There are uncountably more irrational fears than rational ones. -P. Dolan
Liam Forbes loforbes@alaska.edu ph: 907-450-8618 fax: 907-450-8601
UAF Research Computing Systems Senior HPC Engineer CISSP
JR Schmidt
Username: Schmidt

Post Number: 615
Registered: 11-2006
Posted on Thursday, August 02, 2018 - 9:52 am:   

Note that for SLURM you should NOT use qsub. As requested in the 'Batch Queue Manager', provide the path for 'sbatch' (rather than qsub). (The 'qsub' script is just provided for some attempt at compatability with Torque; it is not really for production use.)
Liam Forbes
Unregistered guest
Posted on Thursday, August 02, 2018 - 12:31 pm:   


Ah! Now that you point it out, I see the difference in the web page. I backed out my changes and used the native SLURM commands. It works! Thank you.


Add Your Message Here
Username: Posting Information:
This is a public posting area. Enter your username and password if you have an account. Otherwise, enter your full name as your username and leave the password blank. Your e-mail address is optional.
Options: Post as "Anonymous"
Enable HTML code in message
Automatically activate URLs in message

Topics | Last Day | Last Week | Tree View | Search | Help/Instructions | Program Credits Administration