If you have more than one job to run and you'd like to make sure that your less urgent job doesn't block your more urgent job, try setting 'nice' values:
send your jobs using sbatch <jobscript.sh> and note the job ID then:
sbatch <jobscript.sh>
scontrol update jobid=<jobid> nice=X
And set
A nice job is willing to wait, so higher nice value means lower priority.
A job which is 'less nice' (ie a lower number) will get a higher priority and run sooner.
Thanks to Jamie McKeown for sharing this tip
Back to FAQ index
Table of Contents
HPC Service
Main Content Sections
Documentation Tools