SLURM

This is a user guide, it does not matter if you have or not some experience using Slurm, here you will find some useful examples and commands. We try to cover our most common use cases.

Note

This guide assumes a basic comprehension about Slurm. If you don’t have any idea of what Slurm is or what things it does in the HPC world, we strongly encourage reading some of the articles describe in more information section.

• Submitting jobs
  • What is sbatch?
  • Serial jobs
  • Shared Memory jobs (OpenMP)
  • MPI jobs
  • Array jobs
  • Slurm’s environment variables
  • Slurm’s file-patterns
  • Constraining Features on a job
  • References
• Testing my job
  • Debug partition
  • salloc command
  • --test-only parameter
  • References
• Getting information about jobs
  • Getting cluster(s) state
  • What’s going on with my job? Getting information about submitted jobs
  • Canceling a job
  • What happened with my job? Getting information about finished jobs
  • References
• Fairshare
  • Tested
  • Fair-share Factor
  • Regenerate the initram on CentOS 8
  • Plugins
    • Multifactor Priority Plugin
  • Slurm.conf
  • ¿How do I configure Slurm Fair Shares?
  • More information
  • Authors
• FAQ
  • What should be the CPU Load on my node?
  • In which cases should I use srun?
  • What is the difference between N, n and c?
  • How to know the name(s) of the allocated node(s) inside my job ?

More information

• Quick Start User Guide.
• Slurm + tutorial - Software and Environment Module.
• Morris A. Jette , Andy B. Yoo , Mark Grondona (2002) SLURM: Simple Linux Utility for Resource Management.

Authors

Juan David Arcila-Moreno <jarcil13@eafit.edu.co>