Am 16.12.2015 um 00:37 schrieb Cooper, Trevor:
> If you are the cluster admin and/or have access to the slurm.state directory
> on the slurmctld host the user environment and batch script exist there while
> the job is in the queue.
>
> Sample path would be...
>
> # ls -l /var/spool/slurm.state/hash.0/job.<jobid>
Oh wow! That was easier to find than expected, I just feared before that
"hash.x" would for sure not contain anything human-legible, so I kept looking
in job_state etc. which apparently is something very different.
Thanks a lot for this information, using some 'grep' magic it's now easy to
identify the potentially problematic jobs.
Cheers and thanks again,
Oliver
