Quick reference: Slurm and HPC commands

Introduction

This page lists common Slurm commands and options for jobs. Users familiar with Slurm and job submission on CHTC’s High Performance Computing (HPC) system can use this page as a quick reference. For users who are just starting out, we suggest reading the full guides to understand the full context of each command or option.

Table of Contents

Please note the following!

  • Bracketed items (<>) denote where to place your input. Do not include the brackets in your command.
  • All commands should be entered on a single line.

Submit jobs

See our job submission guide for more information.

Command Use Notes
sbatch <executable.sh> submits job(s) as specified by executable.sh. We recommend using sbatch over the other methods.
srun --mpi=pmix -n4 -N1 -p int --pty bash submits an interactive job (intended for simple testing/compiling) The example uses 4 cores, 1 node, and submits to the interactive partition
scancel <job_ID> removes the job(s) associated with job_ID.  
scontrol hold <job_ID> holds the job(s) associated with job_ID and leaves it in the queue.  
scontrol release <job_ID> releases the held job associated with job_ID to the queue.  

Submit jobs using salloc

We recommend using sbatch to submit jobs (see above) instead of salloc. See this guide for more information.

Command Use Notes
salloc -n4 -N1 -p int requests an allocation for a job The example uses 4 cores, 1 node, and submits to the interactive partition
srun --mpi=pmix </path/to/mpi/script> executes a script using allocated resources (salloc)  
srun --mpi=pmix --pty bash submits an interactive job using allocated resources (salloc)  
echo $SLURM_JOB_ID check to see if you are currently using an allocation. If so, a job ID is returned.  
exit end your allocation  

Monitor jobs

Use these commands to monitor your submitted jobs.

Command Use
squeue --me displays status of your submitted jobs
squeue -u <username> displays status of your submitted jobs by username
squeue -p <partition> displays status of your submitted jobs by partition
seff <job_ID> view job performance

View job information with sacct

sacct displays information about job(s) from the SLURM database. The table below displays options to filter information. Multiple options may be combined.

sacct --jobs <job_ID>,<job_ID> filters by job_ID  
sacct --start=<YYYY-MM-DD> filters jobs starting from date YYYY-MM-DD  
sacct --end=<YYYY-MM-DD> filters jobs ending by date YYYY-MM-DD. Should be used with --start, since the default option only returns jobs from the current day.

Machine information

These commands display information about the machines that execute jobs.

Command Use Notes
sinfo displays information about nodes  
sinfo -n <node> displays information about specified node Example: sinfo -n spark-a006
sinfo -p <partition> displays information about specified partition Example: sinfo -p pre

Modules

These commands access software in the module system. Read this guide for more information.

Command Use
module avail displays available software modules
module spider <keyword> searches for <keyword> in available modules
module load <module> loads <module>
module unload <module> unloads <module>
module list shows all currently loaded modules
module purge unloads all modules

Check quotas

Command Use
get_quotas shows your /home and /scratch disk and files quota
get_quotas -p <path> shows the disk and files quota for a specified path, such as a group directory