Common Issues: Slurm configuration errors

  • Article

Possible Error Messages

  • Unable to execute command /usr/bin/systemctl --system start slurmd

Resolution

Because Slurm requires all of the nodes of a cluster to be defined in slurm.conf, CycleCloud pre-creates all of the VMs inside of CycleCloud when the scheduler node is first started. These VMs remain unallocated in Azure until a job requests them, but sometimes insufficient quota or incorrect autoscale limits can cause installation issues.

  • Make sure your subscription has enough quota for the selected VM type and that the autoscale limits are below the quota amount. You may have to select a different VM type or adjust your autoscale limits accordingly.
  • Check in /var/log/slurmctld/slurm.log for any errors related to starting the scheduler.