1sstat(1) Slurm Commands sstat(1)
2
6 sstat - Display the status information of a running job/step.
7
10 sstat [OPTIONS...]
11
14 Status information for running jobs invoked with Slurm.
15 The sstat command displays job status information for your analysis.
17 The sstat command displays information pertaining to CPU, Task, Node,
18 Resident Set Size (RSS) and Virtual Memory (VM). You can tailor the
19 output with the use of the --fields= option to specify the fields to be
20 shown.
21 For the root user, the sstat command displays job status data for any
23 job running on the system.
24 For the non-root user, the sstat output is limited to the user's jobs.
26 NOTE: The sstat command requires that the jobacct_gather plugin be in‐
29 stalled and operational.
30 NOTE: The sstat command is not supported on Cray ALPS.
32
35 -a, --allsteps
36 Print all steps for the given job(s) when no step is specified.
37 -o, --format, --fields
39 Comma separated list of fields. (use '--helpformat' for a list
40 of available fields).
41 NOTE: When using the format option for listing various fields
43 you can put a %NUMBER afterwards to specify how many characters
44 should be printed.
45 i.e. format=name%30 will print 30 characters of field name right
47 justified. A -30 will print 30 characters left justified.
48 -h, --help
50 Displays a general help message.
51 -e, --helpformat
53 Print a list of fields that can be specified with the '--format'
54 option.
55 -j, --jobs
57 Format is <job(.step)>. Stat this job step or comma-separated
58 list of job steps. This option is required. The step portion
59 will default to the lowest numbered (not batch, extern, etc)
60 step running if not specified, unless the --allsteps flag is set
61 where not specifying a step will result in all running steps to
62 be displayed. NOTE: A step id of 'batch' will display the in‐
63 formation about the batch step. NOTE: A step id of 'extern'
64 will display the information about the extern step. This step
65 is only available when using PrologFlags=contain
66 --noconvert
68 Don't convert units from their original type (e.g. 2048M won't
69 be converted to 2G).
70 -n, --noheader
72 No heading will be added to the output. The default action is to
73 display a header.
74 -p, --parsable
76 output will be '|' delimited with a '|' at the end
77 -P, --parsable2
79 output will be '|' delimited without a '|' at the end
80 -i, --pidformat
82 Predefined format to list the pids running for each job step.
83 (JobId,Nodes,Pids)
84 --usage
86 Display a command usage summary.
87 -v, --verbose
89 Primarily for debugging purposes, report the state of various
90 variables during processing.
91 -V, --version
93 Print version.
94 Job Status Fields
96 Descriptions of each field option can be found below. Note that the
97 Ave*, Max* and Min* accounting fields look at the values for all the
98 tasks of each step in a job and return the average, maximum or minimum
99 values for the job step.
100 AveCPU Average (system + user) CPU time of all tasks in job.
103 AveCPUFreq
105 Average weighted CPU frequency of all tasks in job, in
106 kHz.
107 AveDiskRead
109 Average number of bytes read by all tasks in job.
110 AveDiskWrite
112 Average number of bytes written by all tasks in job.
113 AvePages
115 Average number of page faults of all tasks in job.
116 AveRSS Average resident set size of all tasks in job.
118 AveVMSize
120 Average Virtual Memory size of all tasks in job.
121 ConsumedEnergy
123 Total energy consumed by all tasks in job, in joules.
124 Note: Only in case of exclusive job allocation this value
125 reflects the jobs' real energy consumption.
126 JobID The number of the job or job step. It is in the form:
128 job.jobstep
129 MaxDiskRead
131 Maximum number of bytes read by all tasks in job.
132 MaxDiskReadNode
134 The node on which the maxdiskread occurred.
135 MaxDiskReadTask
137 The task ID where the maxdiskread occurred.
138 MaxDiskWrite
140 Maximum number of bytes written by all tasks in job.
141 MaxDiskWriteNode
143 The node on which the maxdiskwrite occurred.
144 MaxDiskWriteTask
146 The task ID where the maxdiskwrite occurred.
147 MaxPages
149 Maximum number of page faults of all tasks in job.
150 MaxPagesNode
152 The node on which the maxpages occurred.
153 MaxPagesTask
155 The task ID where the maxpages occurred.
156 MaxRSS Maximum resident set size of all tasks in job.
158 MaxRSSNode
160 The node on which the maxrss occurred.
161 MaxRSSTask
163 The task ID where the maxrss occurred.
164 MaxVMSize
166 Maximum Virtual Memory size of all tasks in job.
167 MaxVMSizeNode
169 The node on which the maxvsize occurred.
170 MaxVMSizeTask
172 The task ID where the maxvsize occurred.
173 MinCPU Minimum (system + user) CPU time of all tasks in job.
175 MinCPUNode
177 The node on which the mincpu occurred.
178 MinCPUTask
180 The task ID where the mincpu occurred.
181 NTasks Total number of tasks in a job or step.
183 ReqCPUFreq
185 Requested CPU frequency for the step, in kHz.
186 TresUsageInAve
188 Tres average usage in by all tasks in job. NOTE: If cor‐
189 responding TresUsageInMaxTask is -1 the metric is node
190 centric instead of task.
191 TresUsageInMax
193 Tres maximum usage in by all tasks in job. NOTE: If cor‐
194 responding TresUsageInMaxTask is -1 the metric is node
195 centric instead of task.
196 TresUsageInMaxNode
198 Node for which each maximum TRES usage out occurred.
199 TresUsageInMaxTask
201 Task for which each maximum TRES usage out occurred.
202 TresUsageOutAve
204 Tres average usage out by all tasks in job. NOTE: If
205 corresponding TresUsageOutMaxTask is -1 the metric is
206 node centric instead of task.
207 TresUsageOutMax
209 Tres maximum usage out by all tasks in job. NOTE: If
210 corresponding TresUsageOutMaxTask is -1 the metric is
211 node centric instead of task.
212 TresUsageOutMaxNode
214 Node for which each maximum TRES usage out occurred.
215 TresUsageOutMaxTask
217 Task for which each maximum TRES usage out occurred.
218
220 Executing sstat sends a remote procedure call to slurmctld. If enough
221 calls from sstat or other Slurm client commands that send remote proce‐
222 dure calls to the slurmctld daemon come in at once, it can result in a
223 degradation of performance of the slurmctld daemon, possibly resulting
224 in a denial of service.
225 Do not run sstat or other Slurm client commands that send remote proce‐
227 dure calls to slurmctld from loops in shell scripts or other programs.
228 Ensure that programs limit calls to sstat to the minimum necessary for
229 the information you are trying to gather.
230
233 Some sstat options may be set via environment variables. These environ‐
234 ment variables, along with their corresponding options, are listed be‐
235 low. (Note: Command line options will always override these settings.)
236 SLURM_CONF The location of the Slurm configuration file.
239
241 Display job step information for job 11 with the specified fields:
242 $ sstat --format=AveCPU,AvePages,AveRSS,AveVMSize,JobID -j 11
244 25:02.000 0K 1.37M 5.93M 9.0
245 Display job step information for job 11 with the specified fields in a
248 parsable format:
249 $ sstat -p --format=AveCPU,AvePages,AveRSS,AveVMSize,JobID -j 11
251 25:02.000|0K|1.37M|5.93M|9.0|
252
255 Copyright (C) 2009 Lawrence Livermore National Security. Produced at
256 Lawrence Livermore National Laboratory (cf, DISCLAIMER).
257 Copyright (C) 2010-2022 SchedMD LLC.
258 This file is part of Slurm, a resource management program. For de‐
260 tails, see <https://slurm.schedmd.com/>.
261 Slurm is free software; you can redistribute it and/or modify it under
263 the terms of the GNU General Public License as published by the Free
264 Software Foundation; either version 2 of the License, or (at your op‐
265 tion) any later version.
266 Slurm is distributed in the hope that it will be useful, but WITHOUT
268 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
269 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
270 for more details.
271
274 sacct(1)
275August 2022 Slurm Commands sstat(1)