When on the cluster, you will use the job scheduler to run and interact with your jobs. For the Asha cluster, we use the Slurm Workload Manager. Below is a basic usage outline and links to the official documentation.

• sbatch
  • Used to submit a job that uses a submission script to the cluster. Options for sbatch allow you to:
    • Submit to specific partition.
    • Direct output.
    • Select node count and/or processor count.
    • Change most controls for running your job.
  • More Info
  • Examples.
    • sbatch --partition=all --ntasks 100 example_submission_script.sh
    • sbatch --nodes 2 --job-name=my_job example_submission_script.sh
  • Note: You can set your options in your submission script.
• scancel
  • Cancel a job.
  • Example canceling a specific job.
    • scancel -j 100
  • More Info
• squeue
  • Used to view information on currently running jobs. Using it with no options is often good enough.
  • More Info
  • Note: Can be used with the “watch” command or with the “-i” option to show updated information.
    • Examples
      • watch squeue
      • squeue  -i 1
• sacct
  • Used to view information on past jobs. Use the “-j” option to see a specific job.
  • More Info
• sinfo
  • Get information on partitions in the cluster or specific nodes.
  • Using “sinfo” by itself will show you all the partitions you have access to.
  • sinfo can be manipulated to show only output you would like to see.
    • Example summary of Allocated/Idle/Other(down or in maintenance)/Total cores per partition.
      • sinfo -o "%9P %5a %16C"
  • More Info