sreport

Show cluster utilization (parsable format)

$ sreport -p cluster utilization

$ sreport job sizes printjobcount

$ sreport user topusage

$ sreport cluster AccountUtilizationByUser

$ sreport cluster utilization Start=[YYYY-MM-DD] End=[YYYY-MM-DD]

$ sreport -M [cluster_name] cluster utilization

sreport [OPTIONS...] [COMMAND...]

sreport generates usage and utilization reports from Slurm accounting data stored in slurmdbd. It produces reports on job statistics, user usage, cluster utilization, and reservation usage based on hourly, daily, and monthly rollup tables.
Reports help administrators understand cluster usage patterns, identify top users, and track resource consumption across accounts and projects.

-a, --all_clusters

Report on all clusters, not just local

Specify cluster(s) for reports

Omit header from output

Pipe-delimited output with trailing pipe

Pipe-delimited output without trailing pipe

Set time format for output

Specify trackable resources to report

Report for entire federation

Report local cluster only

Show errors only

Verbose output

Print version

cluster

AccountUtilizationByUser, UserUtilizationByAccount, Utilization, WCKeyUtilizationByUser

SizesByAccount, SizesByAccountAndWckey, SizesByWckey

Utilization

TopUsage

Start date

Report period start (default: previous day 00:00)

Report period end (default: previous day 23:59)

Filter by accounts

Filter by users

Requires slurmdbd to be configured and running. Report accuracy depends on proper accounting configuration. Historical data availability depends on database retention settings. Time periods use cluster local time.

sreport is part of Slurm, developed at Lawrence Livermore National Laboratory starting in 2002. The accounting and reporting features help HPC centers track usage for billing, capacity planning, and fair-share scheduling. Slurm is now maintained by SchedMD.

sacct(1), sshare(1), sacctmgr(1), slurmdbd(8)