Skip to content

profiles/cluster: Only allow Slurm to run when cirrus is mounted

André Breda requested to merge ist189409/nixrnl:cluster-require-cirrus into master

Description of changes

Lab 2 had slurmd running, but /mnt/cirrus wasn't mounted and jobs failed because of it. Why the mount failed I have no idea, but we can at least limit user surprises and have things fail early.

This MR configures systemd to only start slurmd/slurmctld when the cirrus mount succeeds, and to stop slurmd/slurmctld when the mount is also stopped. I am unsure if this also stops slurm services when the mount fails (if that is possible), but hopefully that is the case 😄

Things done

  • Tested
  • Updated documentation (Wiki/NetBox)
  • Breaking change

Merge request reports