Running Jobs in Slurm

Quick Start

A submission script is a shell script that consists of a list of processing tasks that need to be carried out, such as the command, runtime libraries, and input and/or output files for the tasks. If you know the resources that your tasks need to consume, you may also modify the SBATCH script with some of the common directives, e.g.:

Consider a file first.slm

#!/bin/bash
#SBATCH --job-name=MyFirstJob
#SBATCH --time=00:05:00
#SBATCH --ntasks=1
#SBATCH --mem-per-cpu=4096
#SBATCH --cpus-per-task=1
echo "helloworld"

To submit this job, you go

sbatch first.slm

Slurm returns the jobid of the job. You can view the status of the job with:

• scontrol show job <jobid> This show syou the status of pending and running jobs only. See here to viview the status of finished jobs.
• squeue -u $USER This command prints out all the jobs you have.

This script describes the job: it is a serial job with only one process (--ntasks=1). It only needs one CPU core to run the ./helloworld process. The default memory per CPU has been set to 4GB and you should adjust the script based on how much your job needs.

A small list of possible commands

important

Slurm commands (i.e. #SBATCH )must be at the top of your submission scripts. Slurm ignores them once it encounters any non-Slurm commands.

Short Format	Long Format	Default	Description
-N count	--nodes=count	One	Number of nodes to be used. Used to allocate count nodes to your job.
-A accountID	--account=accountID	Your Default	Enter the account ID for your group. You may check your available account(s) with id command.
-t HH:MM:SS	--time=HH:MM:SS	Partition Default	Always specify the maximum wallclock time for your job, max is 7 days.
-p Name	--partition=Name	comp	Specify your partition
-n count	--ntasks	One	Controls the number of tasks (Unix Processes) to be created for the job
N/A	--ntasks-per-node	One	Controls the maximum number of tasks per allocated node
-c count	--cpus-per-task	One	Controls the number of CPUs allocated per task
N/A	--mem-per-cpu	4096MB	Memory size per CPU
-m size	--mem=size	4096MB	Total memory size
-J jobname	--job-name=job_name	slurm-jobid	Up to 15 printable, non-whitespace characters
N/A	--gres=gpu:1	N/A	Generic consumable resources e.g. GPU
N/A	--no-requeue	--requeue	By default, job will be requeued after a node failure

Running Simple Batch Jobs

Submitting a job to SLURM is performed by running the sbatch command and specifying a job script.

sbatch job.script

You can supply options (e.g. --ntasks=xx) to the sbatch command. If an option is already defined in the job.script file, it will be overridden by the commandline argument.

sbatch [options] job.script

Cancelling jobs

To cancel one job

scancel [JOBID]

To cancel all of your jobs

scancel -u [USERID]