Slurm CLI

This page provides a guide for users willing to submit Slurm jobs from the Magellano login shell.

The content of this guide and the examples therein are sufficient for most use cases. However, for mode detailed information, we recommend reading the Slurm User Documentation.

Cluster information (sinfo)

View partition and node information for a system running Slurm.

Usage: sinfo [OPTIONS]

Options

Option

Description

-l, --long

Long output - displays more information

-n, --nodes=NODES

Report on specific node(s)

-N, --Node

Node-centric format

-p,

--partition=PARTITION

Report on specific partition

-s, --summarize

Report state summary only

Examples

Show a summary of available partitions:

$ sinfo -s
PARTITION AVAIL  TIMELIMIT   NODES(A/I/O/T) NODELIST
any*         up 30-00:00:0        1/23/0/24 e1wn[1-20],ne2wn[29-32]
levante      up 30-00:00:0        1/19/0/20 e1wn[1-20]
mefisto      up 30-00:00:0          0/4/0/4 ne2wn[29-32]

Show the status of a subset of nodes in the levante partition:

$ sinfo -N -n e1wn[1-3] -p levante
NODELIST   NODES PARTITION STATE
e1wn1          1   levante mix
e1wn2          1   levante idle
e1wn3          1   levante idle

Batch Jobs (sbatch)

Submit a batch script to Slurm.

Usage: sbatch [OPTIONS...] SCRIPT [ARGS...]

Options

Option

Description

-A,

--account=ACCOUNT

Charge job to specified account, default is default

-c,

--cpus-per-task=NCPUS

Number of cpus required per task, default is 1

-e, --error=ERR

File for batch script’s standard error, default is a file in your current

working directory

-J,

--job-name=JOBNAME

Name of job

--mem=MB

Minimum amount of real memory

--mem-per-cpu=MB

Maximum amount of real memory per allocated cpu required by the job

-n, --ntasks

Number of tasks to run, default is 1

--ntasks-per-node=N

Number of tasks to invoke on each node

-N, --nodes=N

Number of nodes on which to run, or range in the format min[-max]

-o, --output=OUT

File for batch script’s standard output, default is a file in your current

sworking directory

-p,

--partition=PARTITION

Partition requested, default is any

-t, --time

Time limit, default depends on partition configuration

Examples

Submit a non-interactive batch script and require 8 cpus on the mefisto partition:

$ sbatch --cpus-per-task=8 --partition=mefisto my_script.batch

Specify a name for the job, and redirect the standard output and standard error to text files in your home directory:

$ sbatch --job-name=my_job --output=$HOME/my_job.out --error=$HOME/my_job.err my_script.batch

Specify a time limit of 3 hours for your job and charge it to the hpcfriends account:

$ sbatch --account ``hpcfriends`` --time=03:00:00 my_script.batch

Parallel Jobs (srun)

Run parallel jobs.

Usage: srun [OPTIONS...] EXECUTABLE [ARGS...]

Options

Option

Description

-A,

--account=ACCOUNT

Charge job to specified account, default is default

-c,

--cpus-per-task=NCPUS

Number of cpus required per task, default is 1

-e, --error=ERR

File for batch script’s standard error, default is a file in your current

working directory

-J, --job-name=JOBNAME

Name of job

--mem=MB

Minimum amount of real memory

--mem-per-cpu=MB

Maximum amount of real memory per allocated cpu required by the job

-n, --ntasks

Number of tasks to run, default is 1

--ntasks-per-node=N

Number of tasks to invoke on each node

-N, --nodes=N

Number of nodes on which to run, or range in the format min[-max]

-o, --output=OUT

File for batch script’s standard output, default is a file in your current

working directory

-p,

--partition=PARTITION

Partition requested, default is any

--pty

Execute task in pseudo terminal mode (required for parallel shell)

-t, --time

Time limit, default depends on partition configuration

Examples

Run the hostname command in 4 parallel tasks distributed on 4 distinct nodes:

$ srun --ntasks=4 --tasks-per-node=1 hostname -f
e1wn2.pi.infn.it
e1wn1.pi.infn.it
e1wn4.pi.infn.it
e1wn3.pi.infn.it

Start an interactive parallel shell:

$ srun --ntasks=1 --cpus-per-task=4 --pty /bin/bash

Queue Monitoring (squeue)

View information about jobs located in the Slurm scheduling queue.

Usage: squeue [OPTIONS...] EXECUTABLE [ARGS...]

Options

Option

Description

-A,

--account=ACCOUNT(S)

Comma separated list of accounts to view, default is all accounts

-l, --long

Long report

-p,

--partition=PARTITION(S)

Comma separated list of partitions to view, default is all partitions

-u, --user=USER_NAME(S)

Comma separated list of users to view,

Examples

List all queued jobs for user johndoe:

$ squeue --user=johndoe
         JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
          4299   levante s1g2_int  johndoe PD       0:00      1 (ReqNodeNotAvail, UnavailableNodes:e1wn6)
          4259   levante s1g1_int  johndoe PD       0:00      1 (ReqNodeNotAvail, UnavailableNodes:e1wn6)
          4302   levante s1g2_int  johndoe  R    1:08:27      1 e1wn9
          4307   levante s1g2_int  johndoe  R    1:08:27      1 e1wn14
          4262   levante s1g1_int  johndoe  R    1:22:40      1 e1wn9
          4267   levante s1g1_int  johndoe  R    1:22:40      1 e1wn14

List all jobs queued on the levante partition:

$ squeue -p levante
         JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
          4299   levante s1g2_int  johndoe PD       0:00      1 (ReqNodeNotAvail, UnavailableNodes:e1wn6)
          4259   levante s1g1_int  johndoe PD       0:00      1 (ReqNodeNotAvail, UnavailableNodes:e1wn6)
          4101   levante dessis_j      neo  R    5:40:53      1 e1wn1
          4302   levante s1g2_int  johndoe  R    1:10:33      1 e1wn9
          4307   levante s1g2_int  johndoe  R    1:10:33      1 e1wn14
          4262   levante s1g1_int  johndoe  R    1:24:46      1 e1wn9
          4267   levante s1g1_int  johndoe  R    1:24:46      1 e1wn14

Accounting (sacct)

Displays accounting data for all jobs in the Slurm database.

Usage: sacct [OPTIONS...]

Options

Option

Description

-A,

--account=ACCOUNT(S)

Comma separated list of accounts to view, default is all accounts

-E, --endtime=TIME

Select jobs eligible before this time

-j, --jobs=JOB(S)

Display information about this job or comma-separated list of jobs,

default is all jobs

-l, --long

Long report

-r,

--partition=PARTITION(S)

Comma separated list of partitions to select jobs from, default is all

partitions

-S, --starttime=TIME

Select jobs eligible after this time, default is 00:00:00 of the current day

-s, --state=STATE

Select jobs based on their current state or the state they were in

during the time period given

-u, --user=USER_NAME(S)

Comma separated list of users to view,

Hint

Valid start/end time formats are: HH:MM[:SS] [AM|PM], MMDD[YY], MM/DD[/YY], MM.DD[.YY], MM/DD[/YY]-HH:MM[:SS], YYYY-MM-DD[THH:MM[:SS]], now[{+|-}count[seconds(default)|minutes|hours|days|weeks]].

Hint

Valid job states are: running (r), completed (cd), failed (f), timeout (to), resizing (rs), deadline (dl) and node_fail (nf).

Examples

List all jobs of user neo that are in running state:

$ sacct --user=neo --state=running
         JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
          4318   levante dessis_j      neo  R    8:28:47      1 e1wn1
          4319   levante sys/dash      neo  R    1:33:43      1 e1wn1

List all completed jobs in partition levante since January 15:

$ sacct --partition=levante --state=completed --starttime=01.15 --endtime=now
JobID           JobName  Partition    Account  AllocCPUS      State ExitCode
------------ ---------- ---------- ---------- ---------- ---------- --------
3871               bash    levante    default          8  COMPLETED      0:0
3871.extern      extern               default          8  COMPLETED      0:0
3871.0             bash               default          8  COMPLETED      0:0
3954         MATLAB_R2+    levante    default          1  COMPLETED      0:0
3954.batch        batch               default          1  COMPLETED      0:0
3954.extern      extern               default          1  COMPLETED      0:0

Show configuration (scontrol)

View or modify configuration and state of Slurm entities.

Usage: scontrol show [ENTITY=ID]

Examples

Get info about a specific node:

$ scontrol show node e1wn1
NodeName=e1wn1 Arch=x86_64 CoresPerSocket=96
  CPUAlloc=154 CPUEfctv=384 CPUTot=384 CPULoad=1.34
  AvailableFeatures=CPU_MNF:AMD,CPU_GEN:GENOA,CPU_SKU:9654,CPU_FRQ:3.70GHz,IB:NDR
  ActiveFeatures=CPU_MNF:AMD,CPU_GEN:GENOA,CPU_SKU:9654,CPU_FRQ:3.70GHz,IB:NDR
  ...

Get submission options of a job:

$ scontrol show job 12291
JobId=12291 JobName=bash
  UserId=afiori(4136) GroupId=piusers(1144) MCS_label=N/A
  Priority=225 Nice=0 Account=default QOS=low_prio
  JobState=RUNNING Reason=None Dependency=(null)
  Requeue=1 Restarts=0 BatchFlag=0 Reboot=0 ExitCode=0:0
  RunTime=00:34:51 TimeLimit=1-00:00:00 TimeMin=N/A
  SubmitTime=2026-04-27T14:53:09 EligibleTime=2026-04-27T14:53:09
  ...