Submitting jobs on a cluster with MaxArraySize

[YaraToenders]

What happened?
I am trying to submit a job with 188 subjects to a cluster where the MaxArraySize is 120. Following this explanation: https://stackoverflow.com/questions/69543503/start-a-sbatch-array-with-big-number, I added the following lines to slurm.submit.sh:
#SBATCH --array=0-68

VALUES=({121..188})
THISJOBVALUE=${VALUES[$SLURM_TASK_ARRAY_ID]}

singularity run
--contain --cleanenv
--bind /:/ext
/path/to/halfpipe-halfpipe-1.2.2.sif
--workdir /path/to/workdir/
--only-run
--uuid 4f6619b6
--subject-chunks
--only-chunk-index ${THISJOBVALUE}
--nipype-n-procs 2
--keep some

This should avoid going over the max array and still running these subjects. The submission works, however, it seems to only run subject 121 (from the log file:
grep -hnr "Will run subject level chunk" log.txt
826:[2022-09-26 10:44:08,0202] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
840:[2022-09-26 10:44:08,0418] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
848:[2022-09-26 10:44:08,0572] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
849:[2022-09-26 10:44:08,0611] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
850:[2022-09-26 10:44:08,0630] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
851:[2022-09-26 10:44:08,0641] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
1169:[2022-09-26 10:44:08,0662] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
1170:[2022-09-26 10:44:08,0678] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
1171:[2022-09-26 10:44:08,0714] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
1172:[2022-09-26 10:44:08,0718] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
1173:[2022-09-26 10:44:08,0773] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
2450:[2022-09-26 10:44:08,0717] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
2774:[2022-09-26 10:44:08,0602] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
3092:[2022-09-26 10:44:08,0577] [halfpipe ] [INFO ] Will run subject level chunk 121 of 188
)

Do you have any alternative solutions to avoiding the MaxArraySize problem that might work?

What is the exact command that you used to run Halfpipe?
sbatch submit.slurm.sh

Which settings did you use?
Standard ENIGMA settings
log.txt

[HippocampusGirl]

Hi Yara,

Have you tried running with just

 #SBATCH --array=121-188

I can have a more detailed look once I'm back from vacation :-)

[YaraToenders]

Hi Lea,
Thanks! That's what I tried at first but it gives an error when submitting it. The cluster is down today, but I'll post the error tomorrow!

[YaraToenders]

Hi Lea,

The problem is that when using #SBATCH --array=121-188 I am not able to submit the job:

sbatch submit.slurm.sh 
sbatch: error: Batch job submission failed: Invalid job array specification

I had hoped that adding the values would have solved this, but for some reason it doesn't seem to work and I can't figure out why. Thanks again for your time! (and enjoy your holiday!)

[YaraToenders]

Hi Lea,

Have you had a chance to have a look at this?
Thanks!

[HippocampusGirl]

Hi Yara,

It's actually really surprising for me, since this works on our cluster here in Berlin. Would you be available for a short video call to show me what's going on?

[HippocampusGirl]

We figured out the issue! It was a typo, it needs to be SLURM_ARRAY_TASK_ID not SLURM_TASK_ARRAY_ID

[YaraToenders]

Thanks for your help!