[RESOLVED] Qunex generates extraneous slurm files


When running some Qunex container commands I notice that when I use the scheduler extraneous jobs are submitted from the call. For example, if I’m running 5 subjects there will be 9 slurm jobs in my queue. The extraneous ones will error out with errors such as “ERROR: List of sessions missing”. The 5 job files will run normally though. This is mostly a problem because I will often hit the job limit because the command is submitting ~2x as many jobs as it should be (if I run ~80 subjects there will be ~160 jobs submitted) which means I can run half the number of subjects at a time. As a side note - is there are easy implementation of dSQ job submission for qunex so I only submit one job file per session list or will qunex only submit 1 job file per session?

I have run into this problem with many calls but here is one example, run from /gpfs/project/fas/n3/Studies/Connectome/processing/logs/dwiparcellate after I source the qunex environment.

qunex_container dwi_parcellate \
--sessionsfolder="/gpfs/project/fas/n3/Studies/Connectome/subjects" \
--sessions="100206, 100610, 101006, 101309, 101915" \
--matrixversion="3" \
--waytotal="none" \
--parcellationfile="/project/fas/n3/Studies/dmint/DataDrop/Parcellation_Files/Q1-Q6_RelatedValidation210.CorticalAreas_dil_Final_Final_Areas_Group_Colors.32k_fs_LR.dlabel.nii " \
--outname="Glasser_S1200_RelatedValidation210" \
--overwrite="yes" \
--scheduler="SLURM,time=1-00:00:00,ntasks=1,cpus-per-task=1,mem-per-cpu=50000,partition=pi_anticevic_bigmem" \

Slurm files generated for this call are located in :

-rw-rw-r-- 1 ah2252 9579 Jul 30 11:38 slurm-32160792.out
-rw-rw-r-- 1 ah2252 3577 Jul 30 11:38 slurm-32160791.out
-rw-rw-r-- 1 ah2252 9579 Jul 30 11:38 slurm-32160790.out
-rw-rw-r-- 1 ah2252 3577 Jul 30 11:39 slurm-32160793.out
-rw-rw-r-- 1 ah2252 9665 Jul 30 11:39 slurm-32160795.out
-rw-rw-r-- 1 ah2252 9579 Jul 30 11:39 slurm-32160797.out
-rw-rw-r-- 1 ah2252 9665 Jul 30 11:39 slurm-32160796.out
-rw-rw-r-- 1 ah2252 3577 Jul 30 11:39 slurm-32160798.out
-rw-rw-r-- 1 ah2252 3577 Jul 30 11:39 slurm-32160794.out

QuNex expects sessions to be a comma separated list. In your case you have a comma+space separated list. So, instead of --sessions="100206, 100610, 101006, 101309, 101915" it should be --sessions="100206,100610,101006,101309,101915".

I will upgrade this to be more robust for a future release.

1 Like