Page tree
Skip to end of metadata
Go to start of metadata

Use this guide to understand the priority of your jobs in the queue.  A lower priority job will never delay the scheduling of a higher priority job.  Small jobs may squeeze into the gap being created for larger higher priority jobs, if SLURM is certain they will finish before the higher priority job is due to start.

Step-by-step guide

  1. Use the squeue command to display the priority of all jobs in the queue.

    > squeue -o "%8i %8u %15a %.10r %.10L %.5D %.10Q" | less
    JOBID    USER     ACCOUNT             REASON  TIME_LEFT NODES   PRIORITY
    300731   deplazes m72                   None   22:00:13     1       4179
    293572   prey     ga6              Resources 1-00:00:00     8       5406
    296619   tnataraj partner981        Priority 1-00:00:00    35       1498

    In the above output, job 300731 is already running. Job 293572 is higher in the queue than 296619 due to higher priority. Job 293572 will run when resources become available.

  2. Use sprio to determine the components of the priority

    > sprio -j 293572 -l
      JOBID     USER   PRIORITY        AGE  FAIRSHARE    JOBSIZE  PARTITION        QOS   NICE
     293572     prey       5406        430       3971          5       1000          0      0
    > sprio -j 296619 -l
      JOBID     USER   PRIORITY        AGE  FAIRSHARE    JOBSIZE  PARTITION        QOS   NICE
     296619 tnataraj       1498        246        230         23       1000          0      0

    In the above output, the FAIRSHARE is the dominant factor in 293572 having higher priority. The project that this job belongs to has used less of their quarterly allocation so their priority is higher. The AGE column shows that job has been in the queue longer too. Job 296619 has a slight help from having a larger JOBSIZE but it is not a significant factor.