SLURM Commands

SLURM uses command-line commands to control jobs and clusters as well as show detailed information about jobs. The table below presents the most frequently used commands on HPCC. A complete list can be found at the SLURM documentation page. Please also see this SLURM cheatsheet.

Command	Example Usage	Description
srun	srun my_program --arg foo	Run parallel jobs; often used within job scripts.
salloc	salloc -c 2 --time=1:00:00	Request an interactive job on a compute node.
sbatch	sbatch my_job.sb	Used to submit batch jobs to the SLURM scheduler.
squeue	squeue -u user123	View information about jobs - pending or running - in the queue.
scancel	scancel 123456789	Used to cancel jobs or job steps that are under the control of SLURM.
sacct	sacct -o "JobID,AllocCPUS,State"	Display accounting data for all jobs (and job steps) - running, pending, or ended - stored in the SLURM database.
sprio	sprio -j 123456789	View the factors considered for a job's scheduling priority.
sinfo	sinfo -p general-long-gpu	View status information SLURM nodes and partitions.