Essential SLURM commands

sbatch <jobname>
submits the job in file <jobname> to the SLURM queueing system.

scancel <JOBID>
deletes the job with job ID <JOBID> from the SLURM queueing system. Use the squeue command described below to find the job ID of a specific job.

squeue 
prints a list of all waiting and running jobs in the SLURM queueing system. By default the listing includes for all jobs the job ID (JOBID), the SLURM partition, the job name (or at least its first eight characters), the user, the job status (where "PD" indicates a pending and "R" a running job), the time spend while active, the number of nodes requested/actively used for computation, and the list of nodes used when running. In this last column the jobs tagged "Priority" are waiting because higher priority jobs are ahead of them, while the "Resources" tag is given to jobs waiting for the availability of specific resources.

When using the "-u <username>" qualifier only jobs for the user <username> are displayed.

sinfo
checks the status of computing nodes in the SLURM queueing system. For each partition in the SLURM system, the nodes are listed according to their current state (as defined in the STATE column). Typical states are: "idle" (no jobs running), "alloc" (nodes are allocated to jobs that are currently executing), "mix" (the nodes have some of the CPUs allocated, while others are idle), "drain" (the node is unavailable due to an administrative reason), "drng" (the node is running a job, but will after completion not be available due to an administrative reason).