I would like to have all Slurm commands in one place. Here’s a description from Wikipedia about Slurm. The post is incremental and is supplemented with other useful options.

SINFO

Sinfo - is a command that shows information about nodes, e.g. free number, status, partition.

sinfo
    -a, --all (show hidden partitions)
    -d, --dead (show only non-responding nodes nodes)

Slurm node states

  • idle - node ready for allocation
  • mix - node is not fully used
  • alloc - node is fully occupied
  • drain - node is excluded from the queue system
  • down - node is down and unavailable

SQUEUE

Squeue - displays active and pending tasks.

squeue
    -a, --all (show jobs in hidden partitions)
    -u, --user <NAME> (show jobs ordered by the user)

SCANCEL

Scancel - allows you to complete the task.

scancel
    <JOB_PID> (finish the task with the given PID)
    -n, --name <NAME> (finish all tasks with a specific name)
    -u, --user <NAME> (complete all tasks for a specific user)

to be continued…