Add documentation about array jobs
Should include:
- submit syntax (ranges, concurrency, etc)
- using
$SLURM_ARRAY_TASK_ID
- To select dataset (e.g., select a line from a file, a file/directory from a directory listing, etc)
- As random number seed (for reproducible results)
- how it affects emails (generally avoid --mail-type=ARRAY_TASKS for big arrays)
- how to cancel individual id's and the entire array
-
scontrol update jobid=.... arraytaskthrottle=N
to update max concurrent