1scontrol(1) Slurm Commands scontrol(1)
2
3
4
6 scontrol - view or modify Slurm configuration and state.
7
8
10 scontrol [OPTIONS...] [COMMAND...]
11
12
14 scontrol is used to view or modify Slurm configuration including: job,
15 job step, node, partition, reservation, and overall system configura‐
16 tion. Most of the commands can only be executed by user root or an Ad‐
17 ministrator. If an attempt to view or modify configuration information
18 is made by an unauthorized user, an error message will be printed and
19 the requested action will not occur. If no command is entered on the
20 execute line, scontrol will operate in an interactive mode and prompt
21 for input. It will continue prompting for input and executing commands
22 until explicitly terminated. If a command is entered on the execute
23 line, scontrol will execute that command and terminate. All commands
24 and options are case-insensitive, although node names, partition names,
25 and reservation names are case-sensitive (node names "LX" and "lx" are
26 distinct). All commands and options can be abbreviated to the extent
27 that the specification is unique. A modified Slurm configuration can be
28 written to a file using the scontrol write config command. The result‐
29 ing file will be named using the convention "slurm.conf.<datetime>" and
30 located in the same directory as the original "slurm.conf" file. The
31 directory containing the original slurm.conf must be writable for this
32 to occur.
33
34
36 -a, --all
37 When the show command is used, then display all partitions,
38 their jobs and jobs steps. This causes information to be dis‐
39 played about partitions that are configured as hidden and parti‐
40 tions that are unavailable to user's group.
41
42 -M, --clusters=<string>
43 The cluster to issue commands to. Only one cluster name may be
44 specified. Note that the SlurmDBD must be up for this option to
45 work properly. This option implicitly sets the --local option.
46
47 -d, --details
48 Causes the show command to provide additional details where
49 available.
50
51 --federation
52 Report jobs from federation if a member of one.
53
54 -F, --future
55 Report nodes in FUTURE state.
56
57 -h, --help
58 Print a help message describing the usage of scontrol.
59
60 --hide Do not display information about hidden partitions, their jobs
61 and job steps. By default, neither partitions that are config‐
62 ured as hidden nor those partitions unavailable to user's group
63 will be displayed (i.e. this is the default behavior).
64
65 --local
66 Show only information local to this cluster. Ignore other clus‐
67 ters in the federated if a member of one. Overrides --federa‐
68 tion.
69
70 -o, --oneliner
71 Print information one line per record.
72
73 -Q, --quiet
74 Print no warning or informational messages, only fatal error
75 messages.
76
77 --sibling
78 Show all sibling jobs on a federated cluster. Implies --federa‐
79 tion.
80
81 -u, --uid=<uid>
82 Attempt to update a job as user <uid> instead of the invoking
83 user id.
84
85 -v, --verbose
86 Print detailed event logging. Multiple -v's will further in‐
87 crease the verbosity of logging. By default only errors will be
88 displayed.
89
90 -V , --version
91 Print version information and exit.
92
94 abort Instruct the Slurm controller to terminate immediately and gen‐
95 erate a core file. See "man slurmctld" for information about
96 where the core file will be written.
97
98 cancel_reboot <NodeList>
99 Cancel pending reboots on nodes. The node will be undrain'ed and
100 the reason cleared if the node was drained by an ASAP reboot.
101
102 cluster <CLUSTER_NAME>
103 The cluster to issue commands to. Only one cluster name may be
104 specified.
105
106 create <SPECIFICATION>
107 Create a new node, partition, or reservation. See the full list
108 of parameters below.
109
110 completing
111 Display all jobs in a COMPLETING state along with associated
112 nodes in either a COMPLETING or DOWN state.
113
114 delete <SPECIFICATION>
115 Delete the entry with the specified SPECIFICATION. The three
116 SPECIFICATION choices are NodeName=<nodelist>, Partition‐
117 Name=<name> and Reservation=<name>. Nodes can't be deleted if
118 they are in a reservation or have jobs on running. Reservations
119 and partitions should have no associated jobs at the time of
120 their deletion (modify the jobs first). If the specified parti‐
121 tion is in use, the request is denied.
122
123 errnumstr <ERRNO>
124 Given a Slurm error number, return a descriptive string.
125
126 fsdampeningfactor <FACTOR>
127 Set the FairShareDampeningFactor in slurmctld.
128
129 help Display a description of scontrol options and commands.
130
131 hold <job_list>
132 Prevent a pending job from being started (sets its priority to
133 0). Use the release command to permit the job to be scheduled.
134 The job_list argument is a comma separated list of job IDs OR
135 "jobname=" with the job's name, which will attempt to hold all
136 jobs having that name. Note that when a job is held by a system
137 administrator using the hold command, only a system administra‐
138 tor may release the job for execution (also see the uhold com‐
139 mand). When the job is held by its owner, it may also be re‐
140 leased by the job's owner. Additionally, attempting to hold a
141 running job will have not suspend or cancel it. But, it will set
142 the job priority to 0 and update the job reason field, which
143 would hold the job if it was requeued at a later time.
144
145 notify <job_id> <message>
146 Send a message to standard error of the salloc or srun command
147 or batch job associated with the specified job_id.
148
149 pidinfo <proc_id>
150 Print the Slurm job id and scheduled termination time corre‐
151 sponding to the supplied process id, proc_id, on the current
152 node. This will work only with processes on node on which scon‐
153 trol is run, and only for those processes spawned by Slurm and
154 their descendants.
155
156 listpids [<job_id>[.<step_id>]] [<NodeName>]
157 Print a listing of the process IDs in a job step (if JO‐
158 BID.STEPID is provided), or all of the job steps in a job (if
159 job_id is provided), or all of the job steps in all of the jobs
160 on the local node (if job_id is not provided or job_id is "*").
161 This will work only with processes on the node on which scontrol
162 is run, and only for those processes spawned by Slurm and their
163 descendants. Note that some Slurm configurations (ProctrackType
164 value of pgid) are unable to identify all processes associated
165 with a job or job step.
166
167 Note that the NodeName option is only really useful when you
168 have multiple slurmd daemons running on the same host machine.
169 Multiple slurmd daemons on one host are, in general, only used
170 by Slurm developers.
171
172 ping Ping the primary and secondary slurmctld daemon and report if
173 they are responding.
174
175 reboot [ASAP] [nextstate={RESUME|DOWN}] [reason=<reason>] {ALL|<Node‐
176 List>}
177 Reboot the nodes in the system when they become idle using the
178 RebootProgram as configured in Slurm's slurm.conf file. Each
179 node will have the "REBOOT" flag added to its node state. After
180 a node reboots and the slurmd daemon starts up again, the
181 HealthCheckProgram will run once. Then, the slurmd daemon will
182 register itself with the slurmctld daemon and the "REBOOT" flag
183 will be cleared. The node's "DRAIN" state flag will be cleared
184 if the reboot was "ASAP", nextstate=resume or down. The "ASAP"
185 option adds the "DRAIN" flag to each node's state, preventing
186 additional jobs from running on the node so it can be rebooted
187 and returned to service "As Soon As Possible" (i.e. ASAP).
188 "ASAP" will also set the node reason to "Reboot ASAP" if the
189 "reason" option isn't specified. If the "nextstate" option is
190 specified as "DOWN", then the node will remain in a down state
191 after rebooting. If "nextstate" is specified as "RESUME", then
192 the nodes will resume as normal and the node's reason and
193 "DRAIN" state will be cleared. Resuming nodes will be consid‐
194 ered as available in backfill future scheduling and won't be re‐
195 placed by idle nodes in a reservation. The "reason" option sets
196 each node's reason to a user-defined message. A default reason
197 of "reboot requested" is set if no other reason is set on the
198 node. The reason will be appended with: "reboot issued" when
199 the reboot is issued; "reboot complete" when the node registers
200 and has a "nextstate" of "DOWN"; or "reboot timed out" when the
201 node fails to register within ResumeTimeout. You must specify
202 either a list of nodes or that ALL nodes are to be rebooted.
203 NOTE: By default, this command does not prevent additional jobs
204 from being scheduled on any nodes before reboot. To do this,
205 you can either use the "ASAP" option or explicitly drain the
206 nodes beforehand. You can alternately create an advanced reser‐
207 vation to prevent additional jobs from being initiated on nodes
208 to be rebooted. Pending reboots can be cancelled by using
209 "scontrol cancel_reboot <node>" or setting the node state to
210 "CANCEL_REBOOT". A node will be marked "DOWN" if it doesn't re‐
211 boot within ResumeTimeout.
212
213 reconfigure
214 Instruct all Slurm daemons to re-read the configuration file.
215 This command does not restart the daemons. This mechanism can
216 be used to modify configuration parameters set in slurm.conf.
217 The Slurm controller (slurmctld) forwards the request to all
218 other daemons (slurmd daemon on each compute node). Running jobs
219 continue execution. Most configuration parameters can be
220 changed by just running this command; however, there are parame‐
221 ters that require a restart of the relevant Slurm daemons. Pa‐
222 rameters requiring a restart will be noted in the slurm.conf(5)
223 man page. The slurmctld daemon and all slurmd daemons must also
224 be restarted if nodes are added to or removed from the cluster.
225
226 release <job_list>
227 Release a previously held job to begin execution. The job_list
228 argument is a comma separated list of job IDs OR "jobname=" with
229 the job's name, which will attempt to hold all jobs having that
230 name. Also see hold.
231
232 requeue [<option>] <job_list>
233 Requeue a running, suspended or finished Slurm batch job into
234 pending state. The job_list argument is a comma separated list
235 of job IDs. The command accepts the following option:
236
237 Incomplete
238 Operate only on jobs (or tasks of a job array) which have
239 not completed. Specifically only jobs in the following
240 states will be requeued: CONFIGURING, RUNNING, STOPPED or
241 SUSPENDED.
242
243 requeuehold [<option>] <job_list>
244 Requeue a running, suspended or finished Slurm batch job into
245 pending state, moreover the job is put in held state (priority
246 zero). The job_list argument is a comma separated list of job
247 IDs. A held job can be released using scontrol to reset its
248 priority (e.g. "scontrol release <job_id>"). The command ac‐
249 cepts the following options:
250
251 Incomplete
252 Operate only on jobs (or tasks of a job array) which have
253 not completed. Specifically only jobs in the following
254 states will be requeued: CONFIGURING, RUNNING, STOPPED or
255 SUSPENDED.
256
257 State=SpecialExit
258 The "SpecialExit" keyword specifies that the job has to
259 be put in a special state JOB_SPECIAL_EXIT. The "scon‐
260 trol show job" command will display the JobState as SPE‐
261 CIAL_EXIT, while the "squeue" command as SE.
262
263 resume <job_list>
264 Resume a previously suspended job. The job_list argument is a
265 comma separated list of job IDs. Also see suspend.
266
267 NOTE: A suspended job releases its CPUs for allocation to other
268 jobs. Resuming a previously suspended job may result in multi‐
269 ple jobs being allocated the same CPUs, which could trigger gang
270 scheduling with some configurations or severe degradation in
271 performance with other configurations. Use of the scancel com‐
272 mand to send SIGSTOP and SIGCONT signals would stop a job with‐
273 out releasing its CPUs for allocation to other jobs and would be
274 a preferable mechanism in many cases. If performing system
275 maintenance you may want to use suspend/resume in the following
276 way. Before suspending set all nodes to draining or set all par‐
277 titions to down so that no new jobs can be scheduled. Then sus‐
278 pend jobs. Once maintenance is done resume jobs then resume
279 nodes and/or set all partitions back to up. Use with caution.
280
281 schedloglevel <LEVEL>
282 Enable or disable scheduler logging. LEVEL may be "0", "1",
283 "disable" or "enable". "0" has the same effect as "disable". "1"
284 has the same effect as "enable". This value is temporary and
285 will be overwritten when the slurmctld daemon reads the
286 slurm.conf configuration file (e.g. when the daemon is restarted
287 or scontrol reconfigure is executed) if the SlurmSchedLogLevel
288 parameter is present.
289
290 setdebug <LEVEL>
291 Change the debug level of the slurmctld daemon for all active
292 logging channels not originally configured off (quiet). LEVEL
293 may be an integer value between zero and nine (using the same
294 values as SlurmctldDebug in the slurm.conf file) or the name of
295 the most detailed message type to be printed: "quiet", "fatal",
296 "error", "info", "verbose", "debug", "debug2", "debug3", "de‐
297 bug4", or "debug5". This value is temporary and will be over‐
298 written whenever the slurmctld daemon reads the slurm.conf con‐
299 figuration file (e.g. when the daemon is restarted or scontrol
300 reconfigure is executed).
301
302 setdebugflags [+|-]<FLAG>
303 Add or remove DebugFlags of the slurmctld daemon. See "man
304 slurm.conf" for a list of supported DebugFlags.
305
306 NOTE: Changing the value of some DebugFlags will have no effect
307 without restarting the slurmctld daemon, which would set De‐
308 bugFlags based upon the contents of the slurm.conf configuration
309 file or the SLURM_DEBUG_FLAGS environment variable. The environ‐
310 ment variable takes precedence over the setting in the
311 slurm.conf.
312
313 show <ENTITY>[=<ID>] or <ENTITY> [<ID>]
314 Display the state of the specified entity with the specified
315 identification.
316
317 aliases
318 Returns all NodeName values associated with a given Node‐
319 Hostname (useful to get the list of virtual nodes associ‐
320 ated with a real node in a configuration where multiple
321 slurmd daemons execute on a single compute node).
322
323 assoc_mgr
324 Displays the current contents of the slurmctld's internal
325 cache for users, associations and/or qos. The output can
326 be filtered by different record types:
327
328 users=<user1>[...,<userN>]
329 Limit the User Records displayed to those with the
330 specified user name(s).
331
332 accounts=<acct1>[...,<acctN>]
333 Limit the Association Records displayed to those
334 with the specified account name(s).
335
336 qos=<qos1>[...,<qosN>]
337 Limit the QOS Records displayed to those with the
338 specified QOS name(s).
339
340 flags={users|assoc|qos}
341 Specify the desired record type to be displayed.
342 If no flags are specified, all record types are
343 displayed.
344
345 bbstat Displays output from Cray's burst buffer status tool. Op‐
346 tions following bbstat are passed directly to the dwstat
347 command by the slurmctld daemon and the response returned
348 to the user. Equivalent to dwstat.
349
350 burstbuffer
351 Displays the current status of the BurstBuffer plugin.
352
353 config Displays parameter names from the configuration files in
354 mixed case (e.g. SlurmdPort=7003) while derived parame‐
355 ters names are in upper case only (e.g. SLURM_VERSION).
356
357 daemons
358 Reports which daemons should be running on this node.
359
360 dwstat Displays output from Cray's burst buffer status tool. Op‐
361 tions following dwstat are passed directly to the dwstat
362 command by the slurmctld daemon and the response returned
363 to the user. Equivalent to bbstat.
364
365 federation
366 The federation name that the controller is part of and
367 the sibling clusters part of the federation will be
368 listed.
369
370 frontend
371 Shows configured frontend nodes.
372
373 hostlist
374 Takes a list of host names and prints the hostlist ex‐
375 pression for them (the inverse of hostnames). hostlist
376 can also take the absolute pathname of a file (beginning
377 with the character '/') containing a list of hostnames.
378 Multiple node names may be specified using simple node
379 range expressions (e.g. "lx[10-20]"). By default
380 hostlist does not sort the node list or make it unique
381 (e.g. tux2,tux1,tux2 = tux[2,1-2]). If you wanted a
382 sorted list use hostlistsorted (e.g. tux2,tux1,tux2 =
383 tux[1-2,2]).
384
385 hostlistsorted
386 Takes a list of host names and prints a sorted, unique
387 hostlist expression for them. See hostlist.
388
389 hostnames
390 Takes an optional hostlist expression as input and writes
391 a list of individual host names to standard output (one
392 per line). If no hostlist expression is supplied, the
393 contents of the SLURM_JOB_NODELIST environment variable
394 is used. For example "tux[1-3]" is mapped to
395 "tux1","tux2" and "tux3" (one hostname per line).
396
397 job Displays statistics about all jobs by default. If an op‐
398 tional jobid is specified, details for just that job will
399 be displayed. If the job does not specify
400 socket-per-node, cores-per-socket or threads-per-core
401 then it will display '*' in the ReqS:C:T=*:*:* field.
402
403 licenses
404 Displays statistics about all configured licenses (local
405 and remote) by default. If an optional license name is
406 specified, details for just that license will be dis‐
407 played.
408
409
410 node Displays statistics about all nodes by default. If an op‐
411 tional nodename is specified, details for just that node
412 will be displayed.
413
414 partition
415 Displays statistics about all partitions by default. If
416 an optional partition name is specified, details for just
417 that partition will be displayed.
418
419 reservation
420 Displays statistics about all reservations by default. If
421 an optional reservation name is specified, details for
422 just that reservation will be displayed.
423
424 slurmd Displays statistics for the slurmd running on the current
425 node.
426
427 step Displays statistics about all job steps by default. If an
428 optional jobid is specified, details about steps for just
429 that job will be displayed. If a jobid.stepid is speci‐
430 fied, details for just that step will be displayed.
431
432 topology
433 Displays information about the defined topology layout.
434 If a switch is specified, information about that switch
435 will be shown. If one node name is specified, all
436 switches connected to that node (and their parent
437 switches) will be shown. If more than one node name is
438 specified, only switches that connect to all named nodes
439 will be shown.
440
441 shutdown <OPTION>
442 Instruct Slurm daemons to save current state and terminate. By
443 default, the Slurm controller (slurmctld) forwards the request
444 all other daemons (slurmd daemon on each compute node). An OP‐
445 TION of slurmctld or controller results in only the slurmctld
446 daemon being shutdown and the slurmd daemons remaining active.
447
448 suspend <job_list>
449 Suspend a running job. The job_list argument is a comma sepa‐
450 rated list of job IDs. Use the resume command to resume its ex‐
451 ecution. User processes must stop on receipt of SIGSTOP signal
452 and resume upon receipt of SIGCONT for this operation to be ef‐
453 fective. Not all architectures and configurations support job
454 suspension. If a suspended job is requeued, it will be placed
455 in a held state. The time a job is suspended will not count
456 against a job's time limit. Only an operator, administrator,
457 SlurmUser, or root can suspend jobs.
458
459 takeover [<INDEX>]
460 Instruct one of Slurm's backup controllers (slurmctld) to take
461 over system control. By default the first backup controller (IN‐
462 DEX=1) requests control from the primary and waits for its ter‐
463 mination. After that, it switches from backup mode to controller
464 mode. If primary controller can not be contacted, it directly
465 switches to controller mode. This can be used to speed up the
466 Slurm controller fail-over mechanism when the primary node is
467 down. This can be used to minimize disruption if the computer
468 executing the primary Slurm controller is scheduled down.
469 (Note: Slurm's primary controller will take the control back at
470 startup.)
471
472 top <job_list>
473 Move the specified job IDs to the top of the queue of jobs be‐
474 longing to the identical user ID, partition name, account, and
475 QOS. The job_list argument is a comma separated ordered list of
476 job IDs. Any job not matching all of those fields will not be
477 effected. Only jobs submitted to a single partition will be ef‐
478 fected. This operation changes the order of jobs by adjusting
479 job nice values. The net effect on that user's throughput will
480 be negligible to slightly negative. This operation is disabled
481 by default for non-privileged (non-operator, admin, SlurmUser,
482 or root) users. This operation may be enabled for non-privileged
483 users by the system administrator by including the option "en‐
484 able_user_top" in the SchedulerParameters configuration parame‐
485 ter.
486
487 token [lifespan=<lifespan>] [username=<username>]
488 Return an auth token which can be used to support JWT authenti‐
489 cation if AuthAltTypes=auth/jwt has been enabled on the system.
490 Supports two optional arguments. lifespan= may be used to spec‐
491 ify the token's lifespan in seconds. username (only available to
492 SlurmUser/root) may be used to request a token for a different
493 username.
494
495 uhold <job_list>
496 Prevent a pending job from being started (sets its priority to
497 0). The job_list argument is a space separated list of job IDs
498 or job names. Use the release command to permit the job to be
499 scheduled. This command is designed for a system administrator
500 to hold a job so that the job owner may release it rather than
501 requiring the intervention of a system administrator (also see
502 the hold command).
503
504 update <SPECIFICATION>
505 Update job, step, node, partition, or reservation configuration
506 per the supplied specification. SPECIFICATION is in the same
507 format as the Slurm configuration file and the output of the
508 show command described above. It may be desirable to execute the
509 show command (described above) on the specific entity you want
510 to update, then use cut-and-paste tools to enter updated config‐
511 uration values to the update. Note that while most configuration
512 values can be changed using this command, not all can be changed
513 using this mechanism. In particular, the hardware configuration
514 of a node or the physical addition or removal of nodes from the
515 cluster may only be accomplished through editing the Slurm con‐
516 figuration file and executing the reconfigure command (described
517 above).
518
519 version
520 Display the version number of scontrol being executed.
521
522 wait_job <job_id>
523 Wait until a job and all of its nodes are ready for use or the
524 job has entered some termination state. This option is particu‐
525 larly useful in the Slurm Prolog or in the batch script itself
526 if nodes are powered down and restarted automatically as needed.
527
528 NOTE: Don't use scontrol wait_job in PrologSlurmctld or Prolog
529 with PrologFlags=Alloc as this will result in a deadlock.
530
531 NOTE: When using wait_job for an array job, use the SLURM_JOB_ID
532 environment variable to reference the job rather than the
533 SLURM_ARRAY_JOB_ID variable.
534
535
536 write batch_script <job_id> [<optional_filename>]
537 Write the batch script for a given job_id to a file or to std‐
538 out. The file will default to slurm-<job_id>.sh if the optional
539 filename argument is not given. The script will be written to
540 stdout if - is given instead of a filename. The batch script
541 can only be retrieved by an admin or operator, or by the owner
542 of the job.
543
544 write config <optional_filename>
545 Write the current configuration to a file with the naming con‐
546 vention of "slurm.conf.<datetime>" in the same directory as the
547 original slurm.conf file. If a filename is given that file lo‐
548 cation with a .<datetime> suffix is created.
549
551 NOTE: All commands listed below can be used in the interactive mode,
552 but NOT on the initial command line.
553
554
555 all Show all partitions, their jobs and jobs steps. This causes in‐
556 formation to be displayed about partitions that are configured
557 as hidden and partitions that are unavailable to user's group.
558
559 details
560 Causes the show command to provide additional details where
561 available. Job information will include CPUs and NUMA memory
562 allocated on each node. Note that on computers with hyper‐
563 threading enabled and Slurm configured to allocate cores, each
564 listed CPU represents one physical core. Each hyperthread on
565 that core can be allocated a separate task, so a job's CPU count
566 and task count may differ. See the --cpu-bind and --mem-bind
567 option descriptions in srun man pages for more information. The
568 details option is currently only supported for the show job com‐
569 mand.
570
571 exit Terminate scontrol interactive session.
572
573 hide Do not display partition, job or jobs step information for par‐
574 titions that are configured as hidden or partitions that are un‐
575 available to the user's group. This is the default behavior.
576
577 oneliner
578 Print information one line per record.
579
580 quiet Print no warning or informational messages, only fatal error
581 messages.
582
583 quit Terminate the execution of scontrol.
584
585 verbose
586 Print detailed event logging. This includes time-stamps on data
587 structures, record counts, etc.
588
589 !! Repeat the last command executed.
590
592 Note that update requests done by either root, SlurmUser or Administra‐
593 tors are not subject to certain restrictions. For instance, if an Ad‐
594 ministrator changes the QOS on a pending job, certain limits such as
595 the TimeLimit will not be changed automatically as changes made by the
596 Administrators are allowed to violate these restrictions.
597
598
599 Account=<account>
600 Account name to be changed for this job's resource use. Value
601 may be cleared with blank data value, "Account=".
602
603 AdminComment=<spec>
604 Arbitrary descriptive string. Can only be set by a Slurm admin‐
605 istrator.
606
607 ArrayTaskThrottle=<count>
608 Specify the maximum number of tasks in a job array that can exe‐
609 cute at the same time. Set the count to zero in order to elimi‐
610 nate any limit. The task throttle count for a job array is re‐
611 ported as part of its ArrayTaskId field, preceded with a percent
612 sign. For example "ArrayTaskId=1-10%2" indicates the maximum
613 number of running tasks is limited to 2.
614
615 BurstBuffer=<spec>
616 Burst buffer specification to be changed for this job's resource
617 use. Value may be cleared with blank data value, "Burst‐
618 Buffer=". Format is burst buffer plugin specific.
619
620 Clusters=<spec>
621 Specifies the clusters that the federated job can run on.
622
623 ClusterFeatures=<spec>
624 Specifies features that a federated cluster must have to have a
625 sibling job submitted to it. Slurm will attempt to submit a sib‐
626 ling job to a cluster if it has at least one of the specified
627 features.
628
629 Comment=<spec>
630 Arbitrary descriptive string.
631
632 Contiguous={yes|no}
633 Set the job's requirement for contiguous (consecutive) nodes to
634 be allocated. Possible values are "YES" and "NO". Only the
635 Slurm administrator or root can change this parameter.
636
637 CoreSpec=<count>
638 Number of cores to reserve per node for system use. The job
639 will be charged for these cores, but be unable to use them.
640 Will be reported as "*" if not constrained.
641
642 CPUsPerTask=<count>
643 Change the CPUsPerTask job's value.
644
645 Deadline=<time_spec>
646 It accepts times of the form HH:MM:SS to specify a deadline to a
647 job at a specific time of day (seconds are optional). You may
648 also specify midnight, noon, fika (3 PM) or teatime (4 PM) and
649 you can have a time-of-day suffixed with AM or PM for a deadline
650 in the morning or the evening. You can specify a deadline for
651 the job with a date of the form MMDDYY or MM/DD/YY or MM.DD.YY,
652 or a date and time as YYYY-MM-DD[THH:MM[:SS]]. You can also
653 give times like now + count time-units, where the time-units can
654 be seconds (default), minutes, hours, days, or weeks and you can
655 tell Slurm to put a deadline for tomorrow with the keyword to‐
656 morrow. The specified deadline must be later than the current
657 time. Only pending jobs can have the deadline updated. Only
658 the Slurm administrator or root can change this parameter.
659
660 DelayBoot=<time_spec>
661 Change the time to decide whether to reboot nodes in order to
662 satisfy job's feature specification if the job has been eligible
663 to run for less than this time period. See salloc/sbatch man
664 pages option --delay-boot.
665
666 Dependency=<dependency_list>
667 Defer job's initiation until specified job dependency specifica‐
668 tion is satisfied. Cancel dependency with an empty depen‐
669 dency_list (e.g. "Dependency="). <dependency_list> is of the
670 form <type:job_id[:job_id][,type:job_id[:job_id]]>. Many jobs
671 can share the same dependency and these jobs may even belong to
672 different users.
673
674 after:job_id[:jobid...]
675 This job can begin execution after the specified jobs
676 have begun execution or been canceled.
677
678 afterany:job_id[:jobid...]
679 This job can begin execution after the specified jobs
680 have terminated.
681
682 afternotok:job_id[:jobid...]
683 This job can begin execution after the specified jobs
684 have terminated in some failed state (non-zero exit code,
685 node failure, timed out, etc).
686
687 afterok:job_id[:jobid...]
688 This job can begin execution after the specified jobs
689 have successfully executed (ran to completion with an
690 exit code of zero).
691
692 singleton
693 This job can begin execution after any previously
694 launched jobs sharing the same job name and user have
695 terminated. In other words, only one job by that name
696 and owned by that user can be running or suspended at any
697 point in time.
698
699 EligibleTime=<time_spec>
700 See StartTime.
701
702 EndTime
703 The time the job is expected to terminate based on the job's
704 time limit. When the job ends sooner, this field will be up‐
705 dated with the actual end time.
706
707 ExcNodeList=<nodes>
708 Set the job's list of excluded node. Multiple node names may be
709 specified using simple node range expressions (e.g.
710 "lx[10-20]"). Value may be cleared with blank data value, "Exc‐
711 NodeList=".
712
713 Features=<features>
714 Set the job's required node features. The list of features may
715 include multiple feature names separated by ampersand (AND)
716 and/or vertical bar (OR) operators. For example: Fea‐
717 tures="opteron&video" or Features="fast|faster". In the first
718 example, only nodes having both the feature "opteron" AND the
719 feature "video" will be used. There is no mechanism to specify
720 that you want one node with feature "opteron" and another node
721 with feature "video" in case no node has both features. If only
722 one of a set of possible options should be used for all allo‐
723 cated nodes, then use the OR operator and enclose the options
724 within square brackets. For example: "Fea‐
725 tures=[rack1|rack2|rack3|rack4]" might be used to specify that
726 all nodes must be allocated on a single rack of the cluster, but
727 any of those four racks can be used. A request can also specify
728 the number of nodes needed with some feature by appending an as‐
729 terisk and count after the feature name. For example "Fea‐
730 tures=graphics*4" indicates that at least four allocated nodes
731 must have the feature "graphics." Parenthesis are also sup‐
732 ported for features to be ANDed together. For example "Fea‐
733 tures=[(knl&a2a&flat)*4&haswell*2]" indicates the resource allo‐
734 cation should include 4 nodes with ALL of the features "knl",
735 "a2a", and "flat" plus 2 nodes with the feature "haswell". Con‐
736 straints with node counts may only be combined with AND opera‐
737 tors. Value may be cleared with blank data value, for example
738 "Features=".
739
740 Gres=<list>
741 Specifies a comma-delimited list of generic consumable re‐
742 sources. The format of each entry on the list is
743 "name[:count[*cpu]]". The name is that of the consumable re‐
744 source. The count is the number of those resources with a de‐
745 fault value of 1. The specified resources will be allocated to
746 the job on each node allocated unless "*cpu" is appended, in
747 which case the resources will be allocated on a per cpu basis.
748 The available generic consumable resources is configurable by
749 the system administrator. A list of available generic consum‐
750 able resources will be printed and the command will exit if the
751 option argument is "help". Examples of use include
752 "Gres=gpus:2*cpu,disk=40G" and "Gres=help".
753
754 JobId=<job_list>
755 Identify the job(s) to be updated. The job_list may be a comma
756 separated list of job IDs. Either JobId or JobName is required.
757
758 Licenses=<name>
759 Specification of licenses (or other resources available on all
760 nodes of the cluster) as described in salloc/sbatch/srun man
761 pages.
762
763 MailType=<types>
764 Set the mail event types. Valid type values are NONE, BEGIN,
765 END, FAIL, REQUEUE, ALL (equivalent to BEGIN, END, FAIL, RE‐
766 QUEUE, and STAGE_OUT), STAGE_OUT (burst buffer stage out and
767 teardown completed), TIME_LIMIT, TIME_LIMIT_90 (reached 90 per‐
768 cent of time limit), TIME_LIMIT_80 (reached 80 percent of time
769 limit), TIME_LIMIT_50 (reached 50 percent of time limit) and AR‐
770 RAY_TASKS (send emails for each array task). Multiple type val‐
771 ues may be specified in a comma separated list. Unless the AR‐
772 RAY_TASKS option is specified, mail notifications on job BEGIN,
773 END and FAIL apply to a job array as a whole rather than gener‐
774 ating individual email messages for each task in the job array.
775
776 MailUser=<name>
777 Set the user to receive email notification of state changes. A
778 blank string will set the mail user to the default which is the
779 submitting user.
780
781 MinCPUsNode=<count>
782 Set the job's minimum number of CPUs per node to the specified
783 value.
784
785 MinMemoryCPU=<megabytes>
786 Set the job's minimum real memory required per allocated CPU to
787 the specified value. Either MinMemoryCPU or MinMemoryNode may be
788 set, but not both.
789
790 MinMemoryNode=<megabytes>
791 Set the job's minimum real memory required per node to the spec‐
792 ified value. Either MinMemoryCPU or MinMemoryNode may be set,
793 but not both.
794
795 MinTmpDiskNode=<megabytes>
796 Set the job's minimum temporary disk space required per node to
797 the specified value. Only the Slurm administrator or root can
798 change this parameter.
799
800 TimeMin=<timespec>
801 Change TimeMin value which specifies the minimum time limit min‐
802 utes of the job.
803
804 JobName=<name>
805 Identify the name of jobs to be modified or set the job's name
806 to the specified value. When used to identify jobs to be modi‐
807 fied, all jobs belonging to all users are modified unless the
808 UserID option is used to identify a specific user. Either JobId
809 or JobName is required.
810
811 Name[=<name>]
812 See JobName.
813
814 Nice[=<adjustment>]
815 Update the job with an adjusted scheduling priority within
816 Slurm. With no adjustment value the scheduling priority is de‐
817 creased by 100. A negative nice value increases the priority,
818 otherwise decreases it. The adjustment range is +/- 2147483645.
819 Only privileged users can specify a negative adjustment.
820
821 NodeList=<nodes>
822 Change the nodes allocated to a running job to shrink its size.
823 The specified list of nodes must be a subset of the nodes cur‐
824 rently allocated to the job. Multiple node names may be speci‐
825 fied using simple node range expressions (e.g. "lx[10-20]"). Af‐
826 ter a job's allocation is reduced, subsequent srun commands must
827 explicitly specify node and task counts which are valid for the
828 new allocation.
829
830 NumCPUs=<min_count>[-<max_count>]
831 Set the job's minimum and optionally maximum count of CPUs to be
832 allocated.
833
834 NumNodes=<min_count>[-<max_count>]
835 Set the job's minimum and optionally maximum count of nodes to
836 be allocated. If the job is already running, use this to spec‐
837 ify a node count less than currently allocated and resources
838 previously allocated to the job will be relinquished. After a
839 job's allocation is reduced, subsequent srun commands must ex‐
840 plicitly specify node and task counts which are valid for the
841 new allocation. Also see the NodeList parameter above. This is
842 the same as ReqNodes.
843
844 NumTasks=<count>
845 Set the job's count of requested tasks to the specified value.
846 The number of tasks started in a specific step inside the allo‐
847 cation may differ from this value, for instance when a different
848 number of tasks is requested on step creation. This is the same
849 as ReqProcs.
850
851 OverSubscribe={yes|no}
852 Set the job's ability to share compute resources (i.e. individ‐
853 ual CPUs) with other jobs. Possible values are "YES" and "NO".
854 This option can only be changed for pending jobs.
855
856 Partition=<name>
857 Set the job's partition to the specified value.
858
859 Prefer=<features>
860 Set the job's preferred node features. This list is only pre‐
861 ferred, not required like Features is. This list will override
862 what is requested in Features. See Features option above.
863
864 Priority=<number>
865 Set the job's priority to the specified value. Note that a job
866 priority of zero prevents the job from ever being scheduled. By
867 setting a job's priority to zero it is held. Set the priority
868 to a non-zero value to permit it to run. Explicitly setting a
869 job's priority clears any previously set nice value and removes
870 the priority/multifactor plugin's ability to manage a job's pri‐
871 ority. In order to restore the priority/multifactor plugin's
872 ability to manage a job's priority, hold and then release the
873 job. Only the Slurm administrator or root can increase job's
874 priority.
875
876 QOS=<name>
877 Set the job's QOS (Quality Of Service) to the specified value.
878 Value may be cleared with blank data value, "QOS=".
879
880 Reboot={yes|no}
881 Set the job's flag that specifies whether to force the allocated
882 nodes to reboot before starting the job. This is only supported
883 with some system configurations and therefore it could be
884 silently ignored.
885
886 ReqCores=<count>
887 Change the job's requested Cores count.
888
889 ReqNodeList=<nodes>
890 Set the job's list of required node. Multiple node names may be
891 specified using simple node range expressions (e.g.
892 "lx[10-20]"). Value may be cleared with blank data value, "Re‐
893 qNodeList=".
894
895 ReqNodes=<min_count>[-<max_count>]
896 See NumNodes.
897
898 ReqProcs=<count>
899 See NumTasks.
900
901 ReqSockets=<count>
902 Change the job's requested socket count.
903
904 ReqThreads=<count>
905 Change the job's requested threads count.
906
907 IRequeue={0|1}
908 Stipulates whether a job should be requeued after a node fail‐
909 ure: 0 for no, 1 for yes.
910
911 ReservationName=<name>
912 Set the job's reservation to the specified value. Value may be
913 cleared with blank data value, "ReservationName=".
914
915 ResetAccrueTime
916 Set the job's accrue time value to 'now' meaning it will lose
917 any time previously accrued for priority. Helpful if you have a
918 large queue of jobs already in the queue and want to start lim‐
919 iting how many jobs can accrue time without waiting for the
920 queue to flush out.
921
922 SiteFactor=<account>
923 Specify the job's admin priority factor in the range of
924 +/-2147483645. Only privileged users can modify the value.
925
926 StdOut=<filepath>
927 Set the batch job's stdout file path.
928
929 Shared={yes|no}
930 See OverSubscribe option above.
931
932 StartTime=<time_spec>
933 Set the job's earliest initiation time. It accepts times of the
934 form HH:MM:SS to run a job at a specific time of day (seconds
935 are optional). (If that time is already past, the next day is
936 assumed.) You may also specify midnight, noon, fika (3 PM) or
937 teatime (4 PM) and you can have a time-of-day suffixed with AM
938 or PM for running in the morning or the evening. You can also
939 say what day the job will be run, by specifying a date of the
940 form MMDDYY or MM/DD/YY or MM.DD.YY, or a date and time as
941 YYYY-MM-DD[THH:MM[:SS]]. You can also give times like now +
942 count time-units, where the time-units can be seconds (default),
943 minutes, hours, days, or weeks and you can tell Slurm to run the
944 job today with the keyword today and to run the job tomorrow
945 with the keyword tomorrow.
946
947 Notes on date/time specifications:
948 - although the 'seconds' field of the HH:MM:SS time specifica‐
949 tion is allowed by the code, note that the poll time of the
950 Slurm scheduler is not precise enough to guarantee dispatch of
951 the job on the exact second. The job will be eligible to start
952 on the next poll following the specified time. The exact poll
953 interval depends on the Slurm scheduler (e.g., 60 seconds with
954 the default sched/builtin).
955 - if no time (HH:MM:SS) is specified, the default is
956 (00:00:00).
957 - if a date is specified without a year (e.g., MM/DD) then the
958 current year is assumed, unless the combination of MM/DD and
959 HH:MM:SS has already passed for that year, in which case the
960 next year is used.
961
962 Switches=<count>[@<max-time-to-wait>]
963 When a tree topology is used, this defines the maximum count of
964 switches desired for the job allocation. If Slurm finds an allo‐
965 cation containing more switches than the count specified, the
966 job remain pending until it either finds an allocation with de‐
967 sired switch count or the time limit expires. By default there
968 is no switch count limit and no time limit delay. Set the count
969 to zero in order to clean any previously set count (disabling
970 the limit). The job's maximum time delay may be limited by the
971 system administrator using the SchedulerParameters configuration
972 parameter with the max_switch_wait parameter option. Also see
973 wait-for-switch.
974
975 wait-for-switch=<seconds>
976 Change max time to wait for a switch <seconds> secs.
977
978 TasksPerNode=<count>
979 Change the job's requested TasksPerNode.
980
981 ThreadSpec=<count>
982 Number of threads to reserve per node for system use. The job
983 will be charged for these threads, but be unable to use them.
984 Will be reported as "*" if not constrained.
985
986 TimeLimit=<time>
987 The job's time limit. Output format is [days-]hours:min‐
988 utes:seconds or "UNLIMITED". Input format (for update command)
989 set is minutes, minutes:seconds, hours:minutes:seconds,
990 days-hours, days-hours:minutes or days-hours:minutes:seconds.
991 Time resolution is one minute and second values are rounded up
992 to the next minute. If changing the time limit of a job, either
993 specify a new time limit value or precede the time and equal
994 sign with a "+" or "-" to increment or decrement the current
995 time limit (e.g. "TimeLimit+=30"). In order to increment or
996 decrement the current time limit, the JobId specification must
997 precede the TimeLimit specification. Note that incrementing or
998 decrementing the time limit for a job array is only allowed be‐
999 fore the job array has been split into more than one job record.
1000 Only the Slurm administrator or root can increase job's Time‐
1001 Limit.
1002
1003 UserID=<UID or name>
1004 Used with the JobName option to identify jobs to be modified.
1005 Either a user name or numeric ID (UID), may be specified.
1006
1007 WCKey=<key>
1008 Set the job's workload characterization key to the specified
1009 value.
1010
1011 WorkDir=<directory_name>
1012 Set the job's working directory to the specified value. Note
1013 that this may only be set for jobs in the PENDING state, and
1014 that jobs may fail to launch if they rely on relative paths to
1015 the originally submitted WorkDir.
1016
1018 The "show" command, when used with the "job" or "job <jobid>" entity
1019 displays detailed information about a job or jobs. Much of this infor‐
1020 mation may be modified using the "update job" command as described
1021 above. However, the following fields displayed by the show job command
1022 are read-only and cannot be modified:
1023
1024
1025 AllocNode:Sid
1026 Local node and system id making the resource allocation.
1027
1028 BatchFlag
1029 Jobs submitted using the sbatch command have BatchFlag set to 1.
1030 Jobs submitted using other commands have BatchFlag set to 0.
1031
1032 ExitCode=<exit>:<sig>
1033 Exit status reported for the job by the wait() function. The
1034 first number is the exit code, typically as set by the exit()
1035 function. The second number of the signal that caused the
1036 process to terminate if it was terminated by a signal.
1037
1038 GroupId
1039 The group under which the job was submitted.
1040
1041 JobState
1042 The current state of the job.
1043
1044 NodeListIndices
1045 The NodeIndices expose the internal indices into the node table
1046 associated with the node(s) allocated to the job.
1047
1048 NtasksPerN:B:S:C=<tasks_per_node>:<tasks_per_base‐
1049 board>:<tasks_per_socket>:<tasks_per_core>
1050 Specifies the number of tasks to be started per hardware compo‐
1051 nent (node, baseboard, socket and core). Unconstrained values
1052 may be shown as "0" or "*".
1053
1054 PreemptEligibleTime
1055 Time the job becomes eligible for preemption. Modified by Pre‐
1056 emptExemptTime, either from the global option in slurm.conf or
1057 the job QOS. This is hidden if the job has not started or if
1058 PreemptMode=OFF.
1059
1060 PreemptTime
1061 Time at which job was signaled that it was selected for preemp‐
1062 tion. This value is only meaningful for PreemptMode=CANCEL and
1063 PreemptMode=REQUEUE and for jobs in a partition or QOS that has
1064 a GraceTime value designated. This is hidden if the job has not
1065 started or if PreemptMode=OFF.
1066
1067 PreSusTime
1068 Time the job ran prior to last suspend.
1069
1070 Reason The reason job is not running: e.g., waiting "Resources".
1071
1072 ReqB:S:C:T=<baseboard_count>:<socket_per_base‐
1073 board_count>:<core_per_socket_count>:<thread_per_core_count>
1074 Specifies the count of various hardware components requested by
1075 the job. Unconstrained values may be shown as "0" or "*".
1076
1077 SecsPreSuspend=<seconds>
1078 If the job is suspended, this is the run time accumulated by the
1079 job (in seconds) prior to being suspended.
1080
1081 Socks/Node=<count>
1082 Count of desired sockets per node
1083
1084 SubmitTime
1085 The time and date stamp (in localtime) the job was submitted.
1086 The format of the output is identical to that of the EndTime
1087 field.
1088
1089 NOTE: If a job is requeued, the submit time is reset. To obtain
1090 the original submit time it is necessary to use the "sacct -j
1091 <job_id[.<step_id>]" command also designating the -D or --dupli‐
1092 cate option to display all duplicate entries for a job.
1093
1094 SuspendTime
1095 Time the job was last suspended or resumed.
1096
1097 NOTE on information displayed for various job states: When you
1098 submit a request for the "show job" function the scontrol
1099 process makes an RPC request call to slurmctld with a RE‐
1100 QUEST_JOB_INFO message type. If the state of the job is PEND‐
1101 ING, then it returns some detail information such as: min_nodes,
1102 min_procs, cpus_per_task, etc. If the state is other than PEND‐
1103 ING the code assumes that it is in a further state such as RUN‐
1104 NING, COMPLETE, etc. In these cases the code explicitly returns
1105 zero for these values. These values are meaningless once the job
1106 resources have been allocated and the job has started.
1107
1109 StepId=<job_id>[.<step_id>]
1110 Identify the step to be updated. If the job_id is given, but no
1111 step_id is specified then all steps of the identified job will
1112 be modified. This specification is required.
1113
1114 TimeLimit=<time>
1115 The job's time limit. Output format is [days-]hours:min‐
1116 utes:seconds or "UNLIMITED". Input format (for update command)
1117 set is minutes, minutes:seconds, hours:minutes:seconds,
1118 days-hours, days-hours:minutes or days-hours:minutes:seconds.
1119 Time resolution is one minute and second values are rounded up
1120 to the next minute. If changing the time limit of a step, ei‐
1121 ther specify a new time limit value or precede the time with a
1122 "+" or "-" to increment or decrement the current time limit
1123 (e.g. "TimeLimit=+30"). In order to increment or decrement the
1124 current time limit, the StepId specification must precede the
1125 TimeLimit specification.
1126
1128 Provide the same NodeName configuration as found in the slurm.conf. See
1129 slurm.conf man page for details. Only State=CLOUD and State=FUTURE
1130 nodes are allowed.
1131
1132
1134 NodeName=<name>
1135 Identify the node(s) to be updated. Multiple node names may be
1136 specified using simple node range expressions (e.g.
1137 "lx[10-20]"). Nodesets can also be specified by themselves or
1138 mixed with node range expressions, using a comma as a list sepa‐
1139 rator. If the keyword "ALL" is specified alone, then the update
1140 will be attempted against all the nodes in the local cluster.
1141 This specification is required.
1142
1143 ActiveFeatures=<features>
1144 Identify the feature(s) currently active on the specified node.
1145 Any previously active feature specification will be overwritten
1146 with the new value. Also see AvailableFeatures. Typically Ac‐
1147 tiveFeatures will be identical to AvailableFeatures; however Ac‐
1148 tiveFeatures may be configured as a subset of the AvailableFea‐
1149 tures. For example, a node may be booted in multiple configura‐
1150 tions. In that case, all possible configurations may be identi‐
1151 fied as AvailableFeatures, while ActiveFeatures would identify
1152 the current node configuration. When updating the ActiveFea‐
1153 tures with scontrol, the change is only made in slurmctld. When
1154 using a node_features plugin the state/features of the node must
1155 be updated on the node such that a new node start will report
1156 the updated state/features.
1157
1158 AvailableFeatures=<features>
1159 Identify the feature(s) available on the specified node. Any
1160 previously defined available feature specification will be over‐
1161 written with the new value. AvailableFeatures assigned via
1162 scontrol will only persist across the restart of the slurmctld
1163 daemon with the -R option and state files preserved or slurm‐
1164 ctld's receipt of a SIGHUP. Update slurm.conf with any changes
1165 meant to be persistent across normal restarts of slurmctld or
1166 the execution of scontrol reconfig.
1167
1168 Note: Available features being removed via scontrol must not be
1169 active (i.e. remove them from ActiveFeatures first).
1170
1171 Comment=<comment>
1172 Arbitrary descriptive string. Use quotes to enclose a comment
1173 having more than one word
1174
1175 CpuBind=<node>
1176 Specify the task binding mode to be used by default for this
1177 node. Supported options include: "none", "socket", "ldom"
1178 (NUMA), "core", "thread" and "off" (remove previous binding
1179 mode).
1180
1181 Extra=<comment>
1182 Arbitrary string on the node. Use quotes to enclose a string
1183 having more than one word.
1184
1185 Gres=<gres>
1186 Identify generic resources to be associated with the specified
1187 node. Any previously defined generic resources will be over‐
1188 written with the new value. Specifications for multiple generic
1189 resources should be comma separated. Each resource specifica‐
1190 tion consists of a name followed by an optional colon with a nu‐
1191 meric value (default value is one) (e.g. "Gres=band‐
1192 width:10000"). Modification of GRES count associated with spe‐
1193 cific files (e.g. GPUs) is not allowed other than to set their
1194 count on a node to zero. In order to change the GRES count to
1195 another value, modify your slurm.conf and gres.conf files and
1196 restart daemons. If GRES are associated with specific sockets,
1197 that information will be reported For example if all 4 GPUs on a
1198 node are all associated with socket zero, then
1199 "Gres=gpu:4(S:0)". If associated with sockets 0 and 1 then
1200 "Gres=gpu:4(S:0-1)". The information of which specific GPUs are
1201 associated with specific GPUs is not reported, but only avail‐
1202 able by parsing the gres.conf file. Generic resources assigned
1203 via scontrol will only persist across the restart of the slurm‐
1204 ctld daemon with the -R option and state files preserved or
1205 slurmctld's receipt of a SIGHUP. Update slurm.conf with any
1206 changes meant to be persistent across normal restarts of slurm‐
1207 ctld or the execution of scontrol reconfig.
1208
1209 NodeAddr=<node address>
1210 Name that a node should be referred to in establishing a commu‐
1211 nications path. This name will be used as an argument to the
1212 getaddrinfo() function for identification. If a node range ex‐
1213 pression is used to designate multiple nodes, they must exactly
1214 match the entries in the NodeName (e.g. "NodeName=lx[0-7]
1215 NodeAddr=elx[0-7]"). NodeAddr may also contain IP addresses.
1216
1217 NodeHostname=<node hostname>
1218 Typically this would be the string that "/bin/hostname -s" re‐
1219 turns. It may also be the fully qualified domain name as re‐
1220 turned by "/bin/hostname -f" (e.g. "foo1.bar.com"), or any valid
1221 domain name associated with the host through the host database
1222 (/etc/hosts) or DNS, depending on the resolver settings. Note
1223 that if the short form of the hostname is not used, it may pre‐
1224 vent use of hostlist expressions (the numeric portion in brack‐
1225 ets must be at the end of the string). A node range expression
1226 can be used to specify a set of nodes. If an expression is used,
1227 the number of nodes identified by NodeHostname must be identical
1228 to the number of nodes identified by NodeName.
1229
1230 Reason=<reason>
1231 Identify the reason the node is in a "DOWN", "DRAINED", "DRAIN‐
1232 ING", "FAILING" or "FAIL" state. Use quotes to enclose a reason
1233 having more than one word.
1234
1235 State=<state>
1236 Assign one of the following states/actions to the node(s) speci‐
1237 fied by the update command.
1238
1239 CANCEL_REBOOT
1240 Cancels a pending reboot on the node (same as scontrol
1241 cancel_reboot <node>).
1242
1243 DOWN Stop all running and suspended jobs and make the node un‐
1244 available for new jobs.
1245
1246 DRAIN Indicates that no new jobs may be started on this node.
1247 Existing jobs are allowed to run to completion, leaving
1248 the node in a DRAINED state once all the jobs have com‐
1249 pleted.
1250
1251 FAIL Similar to DRAIN except that some applications will seek
1252 to relinquish those nodes before the job completes.
1253
1254 FUTURE Indicates the node is not fully configured, but is ex‐
1255 pected to be available at some point in the future.
1256
1257 NoResp This will set the "Not Responding" flag for a node with‐
1258 out changing its underlying state.
1259
1260 POWER_DOWN
1261 Will use the configured SuspendProgram program to explic‐
1262 itly place a node in power saving mode. If a node is al‐
1263 ready in the process of being powered down, the command
1264 will only change the state of the node but won't have any
1265 effect until the configured SuspendTimeout is reached.
1266 Use of this command can be useful in situations where a
1267 ResumeProgram, like capmc in Cray machines, is stalled
1268 and one wants to restore the node to "IDLE" manually. In
1269 this case rebooting the node and setting the state to
1270 "POWER_DOWN" will cancel the previous "POWER_UP" state
1271 and the node will become "IDLE".
1272
1273 POWER_DOWN_ASAP
1274 Will drain the node and mark it for power down. Currently
1275 running jobs will complete first and no additional jobs
1276 will be allocated to the node.
1277
1278 POWER_DOWN_FORCE
1279 Will cancel all jobs on the node, power it down, and re‐
1280 set its state to "IDLE".
1281
1282 POWER_UP
1283 Will use the configured ResumeProgram program to explic‐
1284 itly move a node out of power saving mode. If a node is
1285 already in the process of being powered up, the command
1286 will only change the state of the node but won't have any
1287 effect until the configured ResumeTimeout is reached.
1288
1289 RESUME Not an actual node state, but will change a node state
1290 from DRAIN, DRAINING, DOWN or REBOOT to IDLE and NoResp.
1291 slurmctld will then attempt to contact slurmd to request
1292 that the node register itself. Once registered, the node
1293 state will then remove the NoResp flag and will resume
1294 normal operations. It will also clear the POWERING_DOWN
1295 state of a node and make it eligible to be allocted.
1296
1297 UNDRAIN
1298 Clears the node from being drained (like RESUME), but
1299 will not change the node's base state (e.g. DOWN).
1300 UNDRAIN requires a valid node registration before new
1301 jobs can be scheduled on the node. Setting a node DOWN
1302 will cause all running and suspended jobs on that node to
1303 be terminated.
1304
1305 While all of the above states are valid, some of them are not
1306 valid new node states given their prior state.
1307
1308 NOTE: The scontrol command should not be used to change node
1309 state on Cray systems. Use Cray tools such as xtprocadmin in‐
1310 stead.
1311
1312 Weight=<weight>
1313 Identify weight to be associated with specified nodes. This al‐
1314 lows dynamic changes to weight associated with nodes, which will
1315 be used for the subsequent node allocation decisions. Weight
1316 assigned via scontrol will only persist across the restart of
1317 the slurmctld daemon with the -R option and state files pre‐
1318 served or slurmctld's receipt of a SIGHUP. Update slurm.conf
1319 with any changes meant to be persistent across normal restarts
1320 of slurmctld or the execution of scontrol reconfig.
1321
1323 NodeName=<nodes>
1324 Identify the node(s) to be deleted. Multiple node names may be
1325 specified using simple node range expressions (e.g.
1326 "lx[10-20]"). Nodesets can also be specified by themselves or
1327 mixed with node range expressions, using a comma as a list sepa‐
1328 rator. If the keyword "ALL" is specified alone, then the update
1329 will be attempted against all the nodes in the local cluster.
1330 This specification is required.
1331
1333 AllocMem
1334 The total memory, in MB, currently allocated by jobs on the
1335 node.
1336
1337 CPULoad
1338 CPU load of a node as reported by the OS.
1339
1340 CPUSpecList
1341 The list of Slurm abstract CPU IDs on this node reserved for ex‐
1342 clusive use by the Slurm compute node daemons (slurmd, slurm‐
1343 stepd).
1344
1345 FreeMem
1346 The total memory, in MB, currently free on the node as reported
1347 by the OS.
1348
1349 LastBusyTime
1350 The last time the node was busy (i.e. last time the node had
1351 jobs on it). This time is used in PowerSave to determine when to
1352 suspend nodes (e.g. now - LastBusy > SuspendTime).
1353
1354 MemSpecLimit
1355 The combined memory limit, in megabytes, on this node for the
1356 Slurm compute node daemons (slurmd, slurmstepd).
1357
1358 RealMemory
1359 The total memory, in MB, on the node.
1360
1361 State Identify the state(s) assigned to the node with '+' delimited
1362 state flags.
1363
1364 States:
1365
1366 ALLOCATED
1367 Indicates that the node has all CPUs allocated to job(s)
1368 running on the node.
1369
1370 DOWN The node does not have any running jobs and is unavail‐
1371 able for new work.
1372
1373 ERROR The node is in an error state. Consult the logs for more
1374 information about what caused this state.
1375
1376 FUTURE The node is currently not fully configured, but expected
1377 to be available at some point in the indefinite future
1378 for use.
1379
1380 IDLE Indicates that the node is available for work but does
1381 not currently have any jobs assigned to it.
1382
1383 MIXED Indicates that the node is in multiple states. For in‐
1384 stance if only part of the node is ALLOCATED and the rest
1385 of the node is IDLE the state will be MIXED.
1386
1387 UNKNOWN
1388 The node has not yet registered with the controller and
1389 its state is not known.
1390
1391 Flags:
1392
1393 CLOUD Indicates that the node is configured as a cloud node, to
1394 be brought up on demand, but not currently running.
1395
1396 COMPLETING
1397 Indicates that the only job on the node or that all jobs
1398 on the node are in the process of completing.
1399
1400 DRAIN The node is not accepting any new jobs and any currently
1401 running jobs will complete.
1402
1403 DYNAMIC
1404 Slurm allows you to define multiple types of nodes in a
1405 FUTURE state. When starting slurmd on a node you can
1406 specify the -F flag to have the node match and use an ex‐
1407 isting definition in your slurm.conf file. The DYNAMIC
1408 state indicates that the node was started as a Dynamic
1409 Future node.
1410
1411 INVALID_REG
1412 The node did not register correctly with the controller.
1413 This happens when a node registers with less resources
1414 than configured in the slurm.conf file. The node will
1415 clear from this state with a valid registration (i.e. a
1416 slurmd restart is required).
1417
1418 MAINTENANCE
1419 The node is currently in a reservation that includes the
1420 maintenance flag.
1421
1422 NOT_RESPONDING
1423 Node is not responding.
1424
1425 PERFCTRS
1426 Indicates that Network Performance Counters associated
1427 with this node are in use, rendering this node as not us‐
1428 able for any other jobs.
1429
1430 POWER_DOWN
1431 Node is pending power down.
1432
1433 POWERED_DOWN
1434 Node is currently powered down and not capable of running
1435 any jobs.
1436
1437 POWERING_DOWN
1438 Node is in the process of powering down.
1439
1440 POWERING_UP
1441 Node is in the process of powering up.
1442
1443 PLANNED
1444 The node is earmarked for a job that will start in the
1445 future.
1446
1447 REBOOT_ISSUED
1448 A reboot request has been sent to the agent configured to
1449 handle this request.
1450
1451 REBOOT_REQUESTED
1452 A request to reboot this node has been made, but hasn't
1453 been handled yet.
1454
1455 RESERVED
1456 Indicates the node is in an advanced reservation and not
1457 generally available.
1458
1459 The meaning of the energy information is as follows:
1460
1461 CurrentWatts
1462 The instantaneous power consumption of the node at the
1463 time of the last node energy accounting sample, in watts.
1464
1465 LowestJoules
1466 The energy consumed by the node between the last time it
1467 was powered on and the last time it was registered by
1468 slurmd, in joules.
1469
1470 ConsumedJoules
1471 The energy consumed by the node between the last time it
1472 was registered by the slurmd daemon and the last node en‐
1473 ergy accounting sample, in joules.
1474
1475 If the reported value is "n/s" (not supported), the node does
1476 not support the configured AcctGatherEnergyType plugin. If the
1477 reported value is zero, energy accounting for nodes is disabled.
1478
1479
1480 The meaning of the external sensors information is as follows:
1481
1482 ExtSensorsJoules
1483 The energy consumed by the node between the last time it
1484 was powered on and the last external sensors plugin node
1485 sample, in joules.
1486
1487 ExtSensorsWatts
1488 The instantaneous power consumption of the node at the
1489 time of the last external sensors plugin node sample, in
1490 watts.
1491
1492 ExtSensorsTemp
1493 The temperature of the node at the time of the last ex‐
1494 ternal sensors plugin node sample, in celsius.
1495
1496 If the reported value is "n/s" (not supported), the node does
1497 not support the configured ExtSensorsType plugin.
1498
1500 FrontendName=<name>
1501 Identify the front end node to be updated. This specification is
1502 required.
1503
1504 Reason=<reason>
1505 Identify the reason the node is in a "DOWN" or "DRAIN" state.
1506 Use quotes to enclose a reason having more than one word.
1507
1508 State=<state>
1509 Identify the state to be assigned to the front end node. Possi‐
1510 ble values are "DOWN", "DRAIN" or "RESUME". If you want to re‐
1511 move a front end node from service, you typically want to set
1512 its state to "DRAIN". "RESUME" is not an actual node state, but
1513 will return a "DRAINED", "DRAINING", or "DOWN" front end node to
1514 service, either "IDLE" or "ALLOCATED" state as appropriate.
1515 Setting a front end node "DOWN" will cause all running and sus‐
1516 pended jobs on that node to be terminated.
1517
1519 AllowGroups=<name>
1520 Identify the user groups which may use this partition. Multiple
1521 groups may be specified in a comma separated list. To permit
1522 all groups to use the partition specify "AllowGroups=ALL".
1523
1524 AllocNodes=<name>
1525 Comma separated list of nodes from which users can execute jobs
1526 in the partition. Node names may be specified using the node
1527 range expression syntax described above. The default value is
1528 "ALL".
1529
1530 Alternate=<partition name>
1531 Alternate partition to be used if the state of this partition is
1532 "DRAIN" or "INACTIVE." The value "NONE" will clear a previously
1533 set alternate partition.
1534
1535 CpuBind=<node>
1536 Specify the task binding mode to be used by default for this
1537 partition. Supported options include: "none", "socket", "ldom"
1538 (NUMA), "core", "thread" and "off" (remove previous binding
1539 mode).
1540
1541 Default={yes|no}
1542 Specify if this partition is to be used by jobs which do not ex‐
1543 plicitly identify a partition to use. Possible output values
1544 are "YES" and "NO". In order to change the default partition of
1545 a running system, use the scontrol update command and set De‐
1546 fault=yes for the partition that you want to become the new de‐
1547 fault.
1548
1549 DefaultTime=<time>
1550 Run time limit used for jobs that don't specify a value. If not
1551 set then MaxTime will be used. Format is the same as for Max‐
1552 Time.
1553
1554 DefMemPerCPU=<MB>
1555 Set the default memory to be allocated per CPU for jobs in this
1556 partition. The memory size is specified in megabytes.
1557
1558 DefMemPerNode=<MB>
1559 Set the default memory to be allocated per node for jobs in this
1560 partition. The memory size is specified in megabytes.
1561
1562 DisableRootJobs={yes|no}
1563 Specify if jobs can be executed as user root. Possible values
1564 are "YES" and "NO".
1565
1566 GraceTime=<seconds>
1567 Specifies, in units of seconds, the preemption grace time to be
1568 extended to a job which has been selected for preemption. The
1569 default value is zero, no preemption grace time is allowed on
1570 this partition or qos. (Meaningful only for PreemptMode=CANCEL)
1571
1572 Hidden={yes|no}
1573 Specify if the partition and its jobs should be hidden from
1574 view. Hidden partitions will by default not be reported by
1575 Slurm APIs or commands. Possible values are "YES" and "NO".
1576
1577 JobDefaults=<specs>
1578 Specify job default values using a comma-delimited list of
1579 "key=value" pairs. Supported keys include
1580
1581 DefCpuPerGPU Default number of CPUs per allocated GPU.
1582
1583 DefMemPerGPU Default memory limit (in megabytes) per allocated
1584 GPU.
1585
1586 MaxMemPerCPU=<MB>
1587 Set the maximum memory to be allocated per CPU for jobs in this
1588 partition. The memory size is specified in megabytes.
1589
1590 MaxMemPerNode=<MB>
1591 Set the maximum memory to be allocated per node for jobs in this
1592 partition. The memory size is specified in megabytes.
1593
1594 MaxNodes=<count>
1595 Set the maximum number of nodes which will be allocated to any
1596 single job in the partition. Specify a number, "INFINITE" or
1597 "UNLIMITED". Changing the MaxNodes of a partition has no effect
1598 upon jobs that have already begun execution.
1599
1600 MaxTime=<time>
1601 The maximum run time for jobs. Output format is
1602 [days-]hours:minutes:seconds or "UNLIMITED". Input format (for
1603 update command) is minutes, minutes:seconds, hours:minutes:sec‐
1604 onds, days-hours, days-hours:minutes or days-hours:minutes:sec‐
1605 onds. Time resolution is one minute and second values are
1606 rounded up to the next minute. Changing the MaxTime of a parti‐
1607 tion has no effect upon jobs that have already begun execution.
1608
1609 MinNodes=<count>
1610 Set the minimum number of nodes which will be allocated to any
1611 single job in the partition. Changing the MinNodes of a parti‐
1612 tion has no effect upon jobs that have already begun execution.
1613 Increasing this value may prevent pending jobs from starting,
1614 even if they were submitted without -N/--nodes specification.
1615 If you do get in that situation, updating the MinNodes value of
1616 a pending job using the scontrol command will allow that job to
1617 be scheduled.
1618
1619 Nodes=<name>
1620 Identify the node(s) to be associated with this partition. Mul‐
1621 tiple node names may be specified using simple node range ex‐
1622 pressions (e.g. "lx[10-20]"). Note that jobs may only be asso‐
1623 ciated with one partition at any time. Specify a blank data
1624 value to remove all nodes from a partition: "Nodes=". Changing
1625 the Nodes in a partition has no effect upon jobs that have al‐
1626 ready begun execution.
1627
1628 OverTimeLimit=<count>
1629 Number of minutes by which a job can exceed its time limit be‐
1630 fore being canceled. The configured job time limit is treated
1631 as a soft limit. Adding OverTimeLimit to the soft limit pro‐
1632 vides a hard limit, at which point the job is canceled. This is
1633 particularly useful for backfill scheduling, which bases upon
1634 each job's soft time limit. A partition-specific OverTimeLimit
1635 will override any global OverTimeLimit value. If not specified,
1636 the global OverTimeLimit value will take precedence. May not
1637 exceed 65533 minutes. An input value of "UNLIMITED" will clear
1638 any previously configured partition-specific OverTimeLimit
1639 value.
1640
1641 OverSubscribe={yes|no|exclusive|force}[:<job_count>]
1642 Specify if compute resources (i.e. individual CPUs) in this par‐
1643 tition can be shared by multiple jobs. Possible values are
1644 "YES", "NO", "EXCLUSIVE" and "FORCE". An optional job count
1645 specifies how many jobs can be allocated to use each resource.
1646
1647 PartitionName=<name>
1648 Identify the partition to be updated. This specification is re‐
1649 quired.
1650
1651 PreemptMode=<mode>
1652 Reset the mechanism used to preempt jobs in this partition if
1653 PreemptType is configured to preempt/partition_prio. The default
1654 preemption mechanism is specified by the cluster-wide Preempt‐
1655 Mode configuration parameter. Possible values are "OFF", "CAN‐
1656 CEL", "REQUEUE" and "SUSPEND".
1657
1658 Priority=<count>
1659 Jobs submitted to a higher priority partition will be dispatched
1660 before pending jobs in lower priority partitions and if possible
1661 they will preempt running jobs from lower priority partitions.
1662 Note that a partition's priority takes precedence over a job's
1663 priority. The value may not exceed 65533.
1664
1665 PriorityJobFactor=<count>
1666 Partition factor used by priority/multifactor plugin in calcu‐
1667 lating job priority. The value may not exceed 65533. Also see
1668 PriorityTier.
1669
1670 PriorityTier=<count>
1671 Jobs submitted to a partition with a higher priority tier value
1672 will be dispatched before pending jobs in partition with lower
1673 priority tier value and, if possible, they will preempt
1674 running jobs from partitions with lower priority tier values.
1675 Note that a partition's priority tier takes precedence over a
1676 job's priority. The value may not exceed 65533. Also see Pri‐
1677 orityJobFactor.
1678
1679 QOS=<QOSname|blank to remove>
1680 Set the partition QOS with a QOS name or to remove the Partition
1681 QOS leave the option blank.
1682
1683 RootOnly={yes|no}
1684 Specify if only allocation requests initiated by user root will
1685 be satisfied. This can be used to restrict control of the par‐
1686 tition to some meta-scheduler. Possible values are "YES" and
1687 "NO".
1688
1689 ReqResv={yes|no}
1690 Specify if only allocation requests designating a reservation
1691 will be satisfied. This is used to restrict partition usage to
1692 be allowed only within a reservation. Possible values are "YES"
1693 and "NO".
1694
1695 Shared={yes|no|exclusive|force}[:<job_count>]
1696 Renamed to OverSubscribe, see option descriptions above.
1697
1698 State={up|down|drain|inactive}
1699 Specify if jobs can be allocated nodes or queued in this parti‐
1700 tion. Possible values are "UP", "DOWN", "DRAIN" and "INACTIVE".
1701
1702 UP Designates that new jobs may queued on the partition,
1703 and that jobs may be allocated nodes and run from the
1704 partition.
1705
1706 DOWN Designates that new jobs may be queued on the parti‐
1707 tion, but queued jobs may not be allocated nodes and
1708 run from the partition. Jobs already running on the
1709 partition continue to run. The jobs must be explicitly
1710 canceled to force their termination.
1711
1712 DRAIN Designates that no new jobs may be queued on the par‐
1713 tition (job submission requests will be denied with an
1714 error message), but jobs already queued on the parti‐
1715 tion may be allocated nodes and run. See also the
1716 "Alternate" partition specification.
1717
1718 INACTIVE Designates that no new jobs may be queued on the par‐
1719 tition, and jobs already queued may not be allocated
1720 nodes and run. See also the "Alternate" partition
1721 specification.
1722
1723 TRESBillingWeights=<TRES Billing Weights>
1724 TRESBillingWeights is used to define the billing weights of each
1725 TRES type that will be used in calculating the usage of a job.
1726 The calculated usage is used when calculating fairshare and when
1727 enforcing the TRES billing limit on jobs. Updates affect new
1728 jobs and not existing jobs. See the slurm.conf man page for
1729 more information.
1730
1732 Reservation=<name>
1733 Identify the name of the reservation to be created, updated, or
1734 deleted. This parameter is required for update and is the only
1735 parameter for delete. For create, if you do not want to give a
1736 reservation name, use "scontrol create reservation ..." and a
1737 name will be created automatically.
1738
1739 Accounts=<account list>
1740 List of accounts permitted to use the reserved nodes, for exam‐
1741 ple "Accounts=physcode1,physcode2". A user in any of the ac‐
1742 counts may use the reserved nodes. A new reservation must spec‐
1743 ify Users or Groups and/or Accounts. If both Users/Groups and
1744 Accounts are specified, a job must match both in order to use
1745 the reservation. Accounts can also be denied access to reserva‐
1746 tions by preceding all of the account names with '-'. Alter‐
1747 nately precede the equal sign with '-'. For example, "Ac‐
1748 counts=-physcode1,-physcode2" or "Accounts-=physcode1,physcode2"
1749 will permit any account except physcode1 and physcode2 to use
1750 the reservation. You can add or remove individual accounts from
1751 an existing reservation by using the update command and adding a
1752 '+' or '-' sign before the '=' sign. If accounts are denied ac‐
1753 cess to a reservation (account name preceded by a '-'), then all
1754 other accounts are implicitly allowed to use the reservation and
1755 it is not possible to also explicitly specify allowed accounts.
1756
1757 BurstBuffer=<buffer_spec>[,<buffer_spec>,...]
1758 Specification of burst buffer resources which are to be re‐
1759 served. "buffer_spec" consists of four elements:
1760 [plugin:][type:]#[units] "plugin" is the burst buffer plugin
1761 name, currently either "datawarp" or "generic". If no plugin is
1762 specified, the reservation applies to all configured burst buf‐
1763 fer plugins. "type" specifies a Cray generic burst buffer re‐
1764 source, for example "nodes". if "type" is not specified, the
1765 number is a measure of storage space. The "units" may be "N"
1766 (nodes), "K|KiB", "M|MiB", "G|GiB", "T|TiB", "P|PiB" (for powers
1767 of 1024) and "KB", "MB", "GB", "TB", "PB" (for powers of 1000).
1768 The default units are bytes for reservations of storage space.
1769 For example "BurstBuffer=datawarp:2TB" (reserve 2TB of storage
1770 plus 3 nodes from the Cray plugin) or "BurstBuffer=100GB" (re‐
1771 serve 100 GB of storage from all configured burst buffer plug‐
1772 ins). Jobs using this reservation are not restricted to these
1773 burst buffer resources, but may use these reserved resources
1774 plus any which are generally available. NOTE: Usually Slurm in‐
1775 terprets KB, MB, GB, TB, PB, TB units as powers of 1024, but for
1776 Burst Buffers size specifications Slurm supports both IEC/SI
1777 formats. This is because the CRAY API for managing DataWarps
1778 supports both formats.
1779
1780 CoreCnt=<num>
1781 This option is only supported when SelectType=select/cons_res or
1782 select/cons_tres. Identify number of cores to be reserved. If
1783 NodeCnt is used without the FIRST_CORES flag, this is the total
1784 number of cores to reserve where cores per node is
1785 CoreCnt/NodeCnt. If a nodelist is used, or if NodeCnt is used
1786 with the FIRST_CORES flag, this should be an array of core num‐
1787 bers by node: Nodes=node[1-5] CoreCnt=2,2,3,3,4 or
1788 flags=FIRST_CORES NodeCnt=5 CoreCnt=1,2,1,3,2.
1789
1790 Licenses=<license>
1791 Specification of licenses (or other resources available on all
1792 nodes of the cluster) which are to be reserved. License names
1793 can be followed by a colon and count (the default count is one).
1794 Multiple license names should be comma separated (e.g. "Li‐
1795 censes=foo:4,bar"). A new reservation must specify one or more
1796 resource to be included: NodeCnt, Nodes and/or Licenses. If a
1797 reservation includes Licenses, but no NodeCnt or Nodes, then the
1798 option Flags=LICENSE_ONLY must also be specified. Jobs using
1799 this reservation are not restricted to these licenses, but may
1800 use these reserved licenses plus any which are generally avail‐
1801 able.
1802
1803 MaxStartDelay[=<timespec>]
1804 Change MaxStartDelay value which specifies the maximum time an
1805 eligible job not requesting this reservation can delay a job re‐
1806 questing it. Default is none. Valid formats are minutes, min‐
1807 utes:seconds, hours:minutes:seconds, days-hours, days-hours:min‐
1808 utes, days-hours:minutes:seconds. Time resolution is one minute
1809 and second values are rounded up to the next minute. Output for‐
1810 mat is always [days-]hours:minutes:seconds.
1811
1812 NodeCnt=<num>[,<num>,...]
1813 Identify number of nodes to be reserved. The number can include
1814 a suffix of "k" or "K", in which case the number specified is
1815 multiplied by 1024. A new reservation must specify one or more
1816 resource to be included: NodeCnt, Nodes and/or Licenses.
1817
1818 Nodes=<name>
1819 Identify the node(s) to be reserved. Multiple node names may be
1820 specified using simple node range expressions (e.g.
1821 "Nodes=lx[10-20]"). When using Nodes to specify more or fewer
1822 nodes, NodeCnt will be updated to honor the new number of nodes.
1823 However, when setting an empty list ("Nodes="), the nodelist
1824 will be filled with random nodes to fulfill the previous nodecnt
1825 and the SPEC_NODES flag will be removed. A new reservation must
1826 specify one or more resource to be included: NodeCnt, Nodes
1827 and/or Licenses. A specification of "ALL" will reserve all
1828 nodes. Set Flags=PART_NODES and PartitionName= in order for
1829 changes in the nodes associated with a partition to also be re‐
1830 flected in the nodes associated with a reservation.
1831
1832 NOTE: When updating a reservation, if Nodes and Nodecnt are set
1833 simultaneously, nodecnt will always be honored. The reservation
1834 will get a subset of nodes if nodes > nodecnt, or it will add
1835 extra nodes to the list when nodes < nodecnt.
1836
1837 StartTime=<time_spec>
1838 The start time for the reservation. A new reservation must
1839 specify a start time. It accepts times of the form HH:MM:SS for
1840 a specific time of day (seconds are optional). (If that time is
1841 already past, the next day is assumed.) You may also specify
1842 midnight, noon, fika (3 PM) or teatime (4 PM) and you can have a
1843 time-of-day suffixed with AM or PM for running in the morning or
1844 the evening. You can also say what day the job will be run, by
1845 specifying a date of the form MMDDYY or MM/DD/YY or MM.DD.YY, or
1846 a date and time as YYYY-MM-DD[THH:MM[:SS]]. You can also give
1847 times like now + count time-units, where the time-units can be
1848 seconds (default), minutes, hours, days, or weeks and you can
1849 tell Slurm to run the job today with the keyword today and to
1850 run the job tomorrow with the keyword tomorrow. You cannot up‐
1851 date the StartTime of a reservation in ACTIVE state.
1852
1853 EndTime=<time_spec>
1854 The end time for the reservation. A new reservation must spec‐
1855 ify an end time or a duration. Valid formats are the same as
1856 for StartTime.
1857
1858 Duration=<time>
1859 The length of a reservation. A new reservation must specify an
1860 end time or a duration. Valid formats are minutes, minutes:sec‐
1861 onds, hours:minutes:seconds, days-hours, days-hours:minutes,
1862 days-hours:minutes:seconds, or UNLIMITED. Time resolution is
1863 one minute and second values are rounded up to the next minute.
1864 Output format is always [days-]hours:minutes:seconds.
1865
1866 PartitionName=<name>
1867 Partition used to reserve nodes from. This will attempt to allo‐
1868 cate all nodes in the specified partition unless you request
1869 fewer resources than are available with CoreCnt, NodeCnt or
1870 TRES. Jobs will be allowed to use this reservation even if run‐
1871 ning in a different partition. There only needs to be overlap‐
1872 ping nodes from that different partition and the nodes used in
1873 the reservation.
1874
1875 Flags=<flags>
1876 Flags associated with the reservation. You can add or remove
1877 individual flags from an existing reservation by adding a '+' or
1878 '-' sign before the '=' sign. For example: Flags-=DAILY (NOTE:
1879 this shortcut is not supported for all flags). Currently sup‐
1880 ported flags include:
1881
1882 ANY_NODES This is a reservation for burst buffers and/or li‐
1883 censes only and not compute nodes. If this flag
1884 is set, a job using this reservation may use the
1885 associated burst buffers and/or licenses plus any
1886 compute nodes. If this flag is not set, a job us‐
1887 ing this reservation may use only the nodes and
1888 licenses associated with the reservation.
1889
1890 DAILY Repeat the reservation at the same time every day.
1891
1892 FLEX Permit jobs requesting the reservation to begin
1893 prior to the reservation's start time, end after
1894 the reservation's end time, and use any resources
1895 inside and/or outside of the reservation regard‐
1896 less of any constraints possibly set in the reser‐
1897 vation. A typical use case is to prevent jobs not
1898 explicitly requesting the reservation from using
1899 those reserved resources rather than forcing jobs
1900 requesting the reservation to use those resources
1901 in the time frame reserved. Another use case could
1902 be to always have a particular number of nodes
1903 with a specific feature reserved for a specific
1904 account so users in this account may use this
1905 nodes plus possibly other nodes without this fea‐
1906 ture.
1907
1908 FIRST_CORES Use the lowest numbered cores on a node only. Flag
1909 removal with '-=' is not supported.
1910
1911 IGNORE_JOBS Ignore currently running jobs when creating the
1912 reservation. This can be especially useful when
1913 reserving all nodes in the system for maintenance.
1914
1915 HOURLY Repeat the reservation at the same time every
1916 hour.
1917
1918 LICENSE_ONLY See ANY_NODES.
1919
1920 MAINT Maintenance mode, receives special accounting
1921 treatment. This reservation is permitted to use
1922 resources that are already in another reservation.
1923
1924 MAGNETIC This flag allows jobs to be considered for this
1925 reservation even if they didn't request it.
1926
1927 NO_HOLD_JOBS_AFTER
1928 By default, when a reservation ends the reserva‐
1929 tion request will be removed from any pending jobs
1930 submitted to the reservation and will be put into
1931 a held state. Use this flag to let jobs run out‐
1932 side of the reservation after the reservation is
1933 gone. Flag removal with '-=' is not supported.
1934
1935 OVERLAP This reservation can be allocated resources that
1936 are already in another reservation. Flag removal
1937 with '-=' is not supported.
1938
1939 PART_NODES This flag can be used to reserve all nodes within
1940 the specified partition. PartitionName and
1941 Nodes=ALL must be specified with this flag.
1942
1943 PURGE_COMP[=<timespec>]
1944 Purge the reservation if it is ever idle for time‐
1945 spec (no jobs associated with it). If timespec
1946 isn't given then 5 minutes is the default. Valid
1947 timespec formats are minutes, minutes:seconds,
1948 hours:minutes:seconds, days-hours, days-hours:min‐
1949 utes, days-hours:minutes:seconds. Time resolution
1950 is one minute and second values are rounded up to
1951 the next minute. Output format is always
1952 [days-]hours:minutes:seconds.
1953
1954 REPLACE Nodes which are DOWN, DRAINED, or allocated to
1955 jobs are automatically replenished using idle re‐
1956 sources. This option can be used to maintain a
1957 constant number of idle resources available for
1958 pending jobs (subject to availability of idle re‐
1959 sources). This should be used with the NodeCnt
1960 reservation option; do not identify specific nodes
1961 to be included in the reservation. Flag removal
1962 with '-=' is not supported.
1963
1964 NOTE: Removing a node from the cluster while in a
1965 reservation with the REPLACE flag will not cause
1966 it to be replaced.
1967
1968 REPLACE_DOWN Nodes which are DOWN or DRAINED are automatically
1969 replenished using idle resources. This option can
1970 be used to maintain a constant sized pool of re‐
1971 sources available for pending jobs (subject to
1972 availability of idle resources). This should be
1973 used with the NodeCnt reservation option; do not
1974 identify specific nodes to be included in the
1975 reservation. Flag removal with '-=' is not sup‐
1976 ported.
1977
1978 NOTE: Removing a node from the cluster while in a
1979 reservation with the REPLACE_DOWN flag will not
1980 cause it to be replaced.
1981
1982 SPEC_NODES Reservation is for specific nodes (output only).
1983
1984 STATIC_ALLOC Make it so after the nodes are selected for a
1985 reservation they don't change. Without this op‐
1986 tion when nodes are selected for a reservation and
1987 one goes down the reservation will select a new
1988 node to fill the spot.
1989
1990 TIME_FLOAT The reservation start time is relative to the cur‐
1991 rent time and moves forward through time (e.g. a
1992 StartTime=now+10minutes will always be 10 minutes
1993 in the future). Repeating (e.g. DAILY) floating
1994 reservations are not supported. Flag cannot be
1995 added to or removed from an existing reservation.
1996
1997 WEEKDAY Repeat the reservation at the same time on every
1998 weekday (Monday, Tuesday, Wednesday, Thursday and
1999 Friday).
2000
2001 WEEKEND Repeat the reservation at the same time on every
2002 weekend day (Saturday and Sunday).
2003
2004 WEEKLY Repeat the reservation at the same time every
2005 week.
2006
2007 Features=<features>
2008 Set the reservation's required node features. Multiple values
2009 may be "&" separated if all features are required (AND opera‐
2010 tion) or separated by "|" if any of the specified features are
2011 required (OR operation). Parenthesis are also supported for
2012 features to be ANDed together with counts of nodes having the
2013 specified features. For example "Fea‐
2014 tures=[(knl&a2a&flat)*4&haswell*2]" indicates the advanced
2015 reservation should include 4 nodes with ALL of the features
2016 "knl", "a2a", and "flat" plus 2 nodes with the feature
2017 "haswell".
2018
2019 Value may be cleared with blank data value, "Features=".
2020
2021 Groups=<group list>
2022 List of groups permitted to use the reserved nodes, for example
2023 "Group=bio,chem". A new reservation must specify Users or
2024 Groups and/or Accounts. If both Users/Groups and Accounts are
2025 specified, a job must match both in order to use the reserva‐
2026 tion. Unlike users groups do not allow denied access to reser‐
2027 vations. You can add or remove individual groups from an exist‐
2028 ing reservation by using the update command and adding a '+' or
2029 '-' sign before the '=' sign. NOTE: Groups and Users are mutu‐
2030 ally exclusive in reservations, if you want to switch between
2031 the 2 you must update the reservation with a group='' or user=''
2032 and fill in the opposite with the appropriate setting.
2033
2034 Skip Used on a reoccurring reservation, skip to the next reservation
2035 iteration. NOTE: Only available for update.
2036
2037 Users=<user list>
2038 List of users permitted to use the reserved nodes, for example
2039 "User=jones1,smith2". A new reservation must specify Users or
2040 Groups and/or Accounts. If both Users/Groups and Accounts are
2041 specified, a job must match both in order to use the reserva‐
2042 tion. Users can also be denied access to reservations by pre‐
2043 ceding all of the user names with '-'. Alternately precede the
2044 equal sign with '-'. For example, "User=-jones1,-smith2" or
2045 "User-=jones1,smith2" will permit any user except jones1 and
2046 smith2 to use the reservation. You can add or remove individual
2047 users from an existing reservation by using the update command
2048 and adding a '+' or '-' sign before the '=' sign. If users are
2049 denied access to a reservation (user name preceded by a '-'),
2050 then all other users are implicitly allowed to use the reserva‐
2051 tion and it is not possible to also explicitly specify allowed
2052 users. NOTE: Groups and Users are mutually exclusive in reser‐
2053 vations, if you want to switch between the 2 you must update the
2054 reservation with a group='' or user='' and fill in the opposite
2055 with the appropriate setting.
2056
2057 TRES=<tres_spec>
2058 Comma-separated list of TRES required for the reservation. Cur‐
2059 rent supported TRES types with reservations are: CPU, Node, Li‐
2060 cense and BB. CPU and Node follow the same format as CoreCnt and
2061 NodeCnt parameters respectively. License names can be followed
2062 by an equal '=' and a count:
2063
2064 License/<name1>=<count1>[,License/<name2>=<count2>,...]
2065
2066 BurstBuffer can be specified in a similar way as BurstBuffer pa‐
2067 rameter. The only difference is that colon symbol ':' should be
2068 replaced by an equal '=' in order to follow the TRES format.
2069
2070 Some examples of TRES valid specifications:
2071
2072 TRES=cpu=5,bb/cray=4,license/iop1=1,license/iop2=3
2073
2074 TRES=node=5k,license/iop1=2
2075
2076 As specified in CoreCnt, if a nodelist is specified, cpu can be
2077 an array of core numbers by node: nodes=compute[1-3]
2078 TRES=cpu=2,2,1,bb/cray=4,license/iop1=2
2079
2080 Please note that CPU, Node, License and BB can override CoreCnt,
2081 NodeCnt, Licenses and BurstBuffer parameters respectively. Also
2082 CPU represents CoreCnt, in a reservation and will be adjusted if
2083 you have threads per core on your nodes.
2084
2085 Note that a reservation that contains nodes or cores is associ‐
2086 ated with one partition, and can't span resources over multiple
2087 partitions. The only exception from this is when the reserva‐
2088 tion is created with explicitly requested nodes.
2089
2091 Executing scontrol sends a remote procedure call to slurmctld. If
2092 enough calls from scontrol or other Slurm client commands that send re‐
2093 mote procedure calls to the slurmctld daemon come in at once, it can
2094 result in a degradation of performance of the slurmctld daemon, possi‐
2095 bly resulting in a denial of service.
2096
2097 Do not run scontrol or other Slurm client commands that send remote
2098 procedure calls to slurmctld from loops in shell scripts or other pro‐
2099 grams. Ensure that programs limit calls to scontrol to the minimum nec‐
2100 essary for the information you are trying to gather.
2101
2102
2104 Some scontrol options may be set via environment variables. These envi‐
2105 ronment variables, along with their corresponding options, are listed
2106 below. (Note: Command line options will always override these set‐
2107 tings.)
2108
2109
2110 SCONTROL_ALL -a, --all
2111
2112 SCONTROL_FEDERATION --federation
2113
2114 SCONTROL_FUTURE -F, --future
2115
2116 SCONTROL_LOCAL --local
2117
2118 SCONTROL_SIBLING --sibling
2119
2120 SLURM_BITSTR_LEN Specifies the string length to be used for holding
2121 a job array's task ID expression. The default
2122 value is 64 bytes. A value of 0 will print the
2123 full expression with any length required. Larger
2124 values may adversely impact the application perfor‐
2125 mance.
2126
2127 SLURM_CLUSTERS Same as --clusters
2128
2129 SLURM_CONF The location of the Slurm configuration file.
2130
2131 SLURM_CONF_OUT When running 'write config', the location of the
2132 Slurm configuration file to be written.
2133
2134 SLURM_DEBUG_FLAGS Specify debug flags for scontrol to use. See De‐
2135 bugFlags in the slurm.conf(5) man page for a full
2136 list of flags. The environment variable takes
2137 precedence over the setting in the slurm.conf.
2138
2139 SLURM_TIME_FORMAT Specify the format used to report time stamps. A
2140 value of standard, the default value, generates
2141 output in the form
2142 "year-month-dateThour:minute:second". A value of
2143 relative returns only "hour:minute:second" if the
2144 current day. For other dates in the current year
2145 it prints the "hour:minute" preceded by "Tomorr"
2146 (tomorrow), "Ystday" (yesterday), the name of the
2147 day for the coming week (e.g. "Mon", "Tue", etc.),
2148 otherwise the date (e.g. "25 Apr"). For other
2149 years it returns a date month and year without a
2150 time (e.g. "6 Jun 2012"). All of the time stamps
2151 use a 24 hour format.
2152
2153 A valid strftime() format can also be specified.
2154 For example, a value of "%a %T" will report the day
2155 of the week and a time stamp (e.g. "Mon 12:34:56").
2156
2157 SLURM_TOPO_LEN Specify the maximum size of the line when printing
2158 Topology. If not set, the default value is unlim‐
2159 ited.
2160
2162 When using SlurmDBD, users who have an AdminLevel defined (Operator or
2163 Admin) and users who are account coordinators are given the authority
2164 to view and modify jobs, reservations, nodes, etc., as defined in the
2165 following table - regardless of whether a PrivateData restriction has
2166 been defined in the slurm.conf file.
2167
2168 scontrol show job(s): Admin, Operator, Coordinator
2169 scontrol update job: Admin, Operator, Coordinator
2170 scontrol requeue: Admin, Operator, Coordinator
2171 scontrol show step(s): Admin, Operator, Coordinator
2172 scontrol update step: Admin, Operator, Coordinator
2173
2174 scontrol show node: Admin, Operator
2175 scontrol update node: Admin
2176
2177 scontrol create partition: Admin
2178 scontrol show partition: Admin, Operator
2179 scontrol update partition: Admin
2180 scontrol delete partition: Admin
2181
2182 scontrol create reservation: Admin, Operator
2183 scontrol show reservation: Admin, Operator
2184 scontrol update reservation: Admin, Operator
2185 scontrol delete reservation: Admin, Operator
2186
2187 scontrol reconfig: Admin
2188 scontrol shutdown: Admin
2189 scontrol takeover: Admin
2190
2191
2193 $ scontrol
2194 scontrol: show part debug
2195 PartitionName=debug
2196 AllocNodes=ALL AllowGroups=ALL Default=YES
2197 DefaultTime=NONE DisableRootJobs=NO Hidden=NO
2198 MaxNodes=UNLIMITED MaxTime=UNLIMITED MinNodes=1
2199 Nodes=snowflake[0-48]
2200 Priority=1 RootOnly=NO OverSubscribe=YES:4
2201 State=UP TotalCPUs=694 TotalNodes=49
2202 scontrol: update PartitionName=debug MaxTime=60:00 MaxNodes=4
2203 scontrol: show job 71701
2204 JobId=71701 Name=hostname
2205 UserId=da(1000) GroupId=da(1000)
2206 Priority=66264 Account=none QOS=normal WCKey=*123
2207 JobState=COMPLETED Reason=None Dependency=(null)
2208 TimeLimit=UNLIMITED Requeue=1 Restarts=0 BatchFlag=0 ExitCode=0:0
2209 SubmitTime=2010-01-05T10:58:40 EligibleTime=2010-01-05T10:58:40
2210 StartTime=2010-01-05T10:58:40 EndTime=2010-01-05T10:58:40
2211 SuspendTime=None SecsPreSuspend=0
2212 Partition=debug AllocNode:Sid=snowflake:4702
2213 ReqNodeList=(null) ExcNodeList=(null)
2214 NodeList=snowflake0
2215 NumNodes=1 NumCPUs=10 CPUs/Task=2 ReqS:C:T=1:1:1
2216 MinCPUsNode=2 MinMemoryNode=0 MinTmpDiskNode=0
2217 Features=(null) Reservation=(null)
2218 OverSubscribe=OK Contiguous=0 Licenses=(null) Network=(null)
2219 scontrol: update JobId=71701 TimeLimit=30:00 Priority=500
2220 scontrol: show hostnames tux[1-3]
2221 tux1
2222 tux2
2223 tux3
2224 scontrol: create res StartTime=2009-04-01T08:00:00 Duration=5:00:00 Users=dbremer NodeCnt=10
2225 Reservation created: dbremer_1
2226 scontrol: update Reservation=dbremer_1 Flags=Maint NodeCnt=20
2227 scontrol: delete Reservation=dbremer_1
2228 scontrol: quit
2229
2230
2232 Copyright (C) 2002-2007 The Regents of the University of California.
2233 Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
2234 Copyright (C) 2008-2010 Lawrence Livermore National Security.
2235 Copyright (C) 2010-2022 SchedMD LLC.
2236
2237 This file is part of Slurm, a resource management program. For de‐
2238 tails, see <https://slurm.schedmd.com/>.
2239
2240 Slurm is free software; you can redistribute it and/or modify it under
2241 the terms of the GNU General Public License as published by the Free
2242 Software Foundation; either version 2 of the License, or (at your op‐
2243 tion) any later version.
2244
2245 Slurm is distributed in the hope that it will be useful, but WITHOUT
2246 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
2247 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
2248 for more details.
2249
2251 /etc/slurm.conf
2252
2254 scancel(1), sinfo(1), squeue(1), slurm_create_partition [22m(3),
2255 slurm_delete_partition [22m(3), slurm_load_ctl_conf [22m(3), slurm_load_jobs
2256 (3), slurm_load_node (3), slurm_load_partitions (3), slurm_reconfigure
2257 (3), slurm_requeue [22m(3), slurm_resume [22m(3), slurm_shutdown [22m(3),
2258 slurm_suspend (3), slurm_takeover (3), slurm_update_job (3), slurm_up‐
2259 date_node (3), slurm_update_partition (3), slurm.conf(5), slurmctld(8)
2260
2261
2262
2263November 2022 Slurm Commands scontrol(1)