File access problem on PBS cluster

Dear community,

I am trying to run Tutorial 1 on a PBS cluster. The jobs can be submitted successfully to the server, but they got killed immediately. The error messages are like this:

/var/spool/PBS/mom_priv/jobs/8588204.pbs.SC: line 63: checkjob: command not found
/var/spool/PBS/mom_priv/jobs/8588204.pbs.SC: line 64: qstat: command not found
grep: …/workflow/control/all.ctrl: No such file or directory
grep: …/workflow/control/all.ctrl: No such file or directory
mkdir: cannot create directory ‘/tangm5’: Permission denied

Our IT team commented as follows:

“The file you are trying to access in the mom_priv area is actually located on the nodes and is protected for the pbs daemon. Its basically the submit script copied over to actually do the run.
I think you might have other issues in your program. The reason being that for one the output file generated is called job-1.1_${PBS_JOBID}.out and not the resolved PBS_JOBID. Also it looks like your PATH variable is not honoured and that is why the qstat commands can’t be found.”

Does anyone have any suggestions for my issues?


Dear Ming,

Welcome to the forum :slight_smile:

It sounds like you are using a PBS cluster on which the commands and environment variables which are used in the job template are not present. On the PBS/Moab cluster which we used, these commands/variables were present. See also here in the documentation:

You might need to adjust the job template according to your specific PBS cluster. Your admins might be able to help with that. Alternatively, if you have access to a Slurm cluster, you might be able to avoid such custom modifications and use the Slurm job templates instead. PBS is not well maintained anymore, and Slurm is much more modern and commonly used.


Dear Christoph,

Thank you very much for your guidance. My problem has been solved, and it is working on our PBS HPC. Virtual Flow is an amasing program.


1 Like