1srun(1) Slurm Commands srun(1)
2
6 srun - Run parallel jobs
7
10 srun [OPTIONS(0)...] [ : [OPTIONS(N)...]] executable(0) [args(0)...]
11 Option(s) define multiple jobs in a co-scheduled heterogeneous job.
13 For more details about heterogeneous jobs see the document
14 https://slurm.schedmd.com/heterogeneous_jobs.html
15
18 Run a parallel job on cluster managed by Slurm. If necessary, srun
19 will first create a resource allocation in which to run the parallel
20 job.
21 The following document describes the influence of various options on
23 the allocation of cpus to jobs and tasks.
24 https://slurm.schedmd.com/cpu_management.html
25
28 srun will return the highest exit code of all tasks run or the highest
29 signal (with the high-order bit set in an 8-bit integer -- e.g. 128 +
30 signal) of any task that exited with a signal.
31
34 The executable is resolved in the following order:
35 1. If executable starts with ".", then path is constructed as: current
37 working directory / executable
38 2. If executable starts with a "/", then path is considered absolute.
40 3. If executable can be resolved through PATH. See path_resolution(7).
42 4. If executable is in current working directory.
44 Current working directory is the calling process working directory
46 unless the --chdir argument is passed, which will override the current
47 working directory.
48
51 --accel-bind=<options>
52 Control how tasks are bound to generic resources of type gpu,
53 mic and nic. Multiple options may be specified. Supported
54 options include:
55 g Bind each task to GPUs which are closest to the allocated
57 CPUs.
58 m Bind each task to MICs which are closest to the allocated
60 CPUs.
61 n Bind each task to NICs which are closest to the allocated
63 CPUs.
64 v Verbose mode. Log how tasks are bound to GPU and NIC
66 devices.
67 This option applies to job allocations.
69 -A, --account=<account>
72 Charge resources used by this job to specified account. The
73 account is an arbitrary string. The account name may be changed
74 after job submission using the scontrol command. This option
75 applies to job allocations.
76 --acctg-freq
79 Define the job accounting and profiling sampling intervals.
80 This can be used to override the JobAcctGatherFrequency parame‐
81 ter in Slurm's configuration file, slurm.conf. The supported
82 format is follows:
83 --acctg-freq=<datatype>=<interval>
85 where <datatype>=<interval> specifies the task sam‐
86 pling interval for the jobacct_gather plugin or a
87 sampling interval for a profiling type by the
88 acct_gather_profile plugin. Multiple, comma-sepa‐
89 rated <datatype>=<interval> intervals may be speci‐
90 fied. Supported datatypes are as follows:
91 task=<interval>
93 where <interval> is the task sampling inter‐
94 val in seconds for the jobacct_gather plugins
95 and for task profiling by the
96 acct_gather_profile plugin. NOTE: This fre‐
97 quency is used to monitor memory usage. If
98 memory limits are enforced the highest fre‐
99 quency a user can request is what is config‐
100 ured in the slurm.conf file. They can not
101 turn it off (=0) either.
102 energy=<interval>
104 where <interval> is the sampling interval in
105 seconds for energy profiling using the
106 acct_gather_energy plugin
107 network=<interval>
109 where <interval> is the sampling interval in
110 seconds for infiniband profiling using the
111 acct_gather_infiniband plugin.
112 filesystem=<interval>
114 where <interval> is the sampling interval in
115 seconds for filesystem profiling using the
116 acct_gather_filesystem plugin.
117 The default value for the task sampling
119 interval
120 is 30. The default value for all other intervals is 0. An
121 interval of 0 disables sampling of the specified type. If the
122 task sampling interval is 0, accounting information is collected
123 only at job termination (reducing Slurm interference with the
124 job).
125 Smaller (non-zero) values have a greater impact upon job perfor‐
126 mance, but a value of 30 seconds is not likely to be noticeable
127 for applications having less than 10,000 tasks. This option
128 applies job allocations.
129 -B --extra-node-info=<sockets[:cores[:threads]]>
132 Restrict node selection to nodes with at least the specified
133 number of sockets, cores per socket and/or threads per core.
134 NOTE: These options do not specify the resource allocation size.
135 Each value specified is considered a minimum. An asterisk (*)
136 can be used as a placeholder indicating that all available
137 resources of that type are to be utilized. Values can also be
138 specified as min-max. The individual levels can also be speci‐
139 fied in separate options if desired:
140 --sockets-per-node=<sockets>
141 --cores-per-socket=<cores>
142 --threads-per-core=<threads>
143 If task/affinity plugin is enabled, then specifying an alloca‐
144 tion in this manner also sets a default --cpu-bind option of
145 threads if the -B option specifies a thread count, otherwise an
146 option of cores if a core count is specified, otherwise an
147 option of sockets. If SelectType is configured to
148 select/cons_res, it must have a parameter of CR_Core,
149 CR_Core_Memory, CR_Socket, or CR_Socket_Memory for this option
150 to be honored. If not specified, the scontrol show job will
151 display 'ReqS:C:T=*:*:*'. This option applies to job alloca‐
152 tions.
153 --bb=<spec>
156 Burst buffer specification. The form of the specification is
157 system dependent. Also see --bbf. This option applies to job
158 allocations.
159 --bbf=<file_name>
162 Path of file containing burst buffer specification. The form of
163 the specification is system dependent. Also see --bb. This
164 option applies to job allocations.
165 --bcast[=<dest_path>]
168 Copy executable file to allocated compute nodes. If a file name
169 is specified, copy the executable to the specified destination
170 file path. If no path is specified, copy the file to a file
171 named "slurm_bcast_<job_id>.<step_id>" in the current working.
172 For example, "srun --bcast=/tmp/mine -N3 a.out" will copy the
173 file "a.out" from your current directory to the file "/tmp/mine"
174 on each of the three allocated compute nodes and execute that
175 file. This option applies to step allocations.
176 -b, --begin=<time>
179 Defer initiation of this job until the specified time. It
180 accepts times of the form HH:MM:SS to run a job at a specific
181 time of day (seconds are optional). (If that time is already
182 past, the next day is assumed.) You may also specify midnight,
183 noon, fika (3 PM) or teatime (4 PM) and you can have a
184 time-of-day suffixed with AM or PM for running in the morning or
185 the evening. You can also say what day the job will be run, by
186 specifying a date of the form MMDDYY or MM/DD/YY YYYY-MM-DD.
187 Combine date and time using the following format
188 YYYY-MM-DD[THH:MM[:SS]]. You can also give times like now +
189 count time-units, where the time-units can be seconds (default),
190 minutes, hours, days, or weeks and you can tell Slurm to run the
191 job today with the keyword today and to run the job tomorrow
192 with the keyword tomorrow. The value may be changed after job
193 submission using the scontrol command. For example:
194 --begin=16:00
195 --begin=now+1hour
196 --begin=now+60 (seconds by default)
197 --begin=2010-01-20T12:34:00
198 Notes on date/time specifications:
201 - Although the 'seconds' field of the HH:MM:SS time specifica‐
202 tion is allowed by the code, note that the poll time of the
203 Slurm scheduler is not precise enough to guarantee dispatch of
204 the job on the exact second. The job will be eligible to start
205 on the next poll following the specified time. The exact poll
206 interval depends on the Slurm scheduler (e.g., 60 seconds with
207 the default sched/builtin).
208 - If no time (HH:MM:SS) is specified, the default is
209 (00:00:00).
210 - If a date is specified without a year (e.g., MM/DD) then the
211 current year is assumed, unless the combination of MM/DD and
212 HH:MM:SS has already passed for that year, in which case the
213 next year is used.
214 This option applies to job allocations.
215 --checkpoint=<time>
218 Specifies the interval between creating checkpoints of the job
219 step. By default, the job step will have no checkpoints cre‐
220 ated. Acceptable time formats include "minutes", "minutes:sec‐
221 onds", "hours:minutes:seconds", "days-hours", "days-hours:min‐
222 utes" and "days-hours:minutes:seconds". This option applies to
223 job and step allocations.
224 --cluster-constraint=<list>
227 Specifies features that a federated cluster must have to have a
228 sibling job submitted to it. Slurm will attempt to submit a sib‐
229 ling job to a cluster if it has at least one of the specified
230 features.
231 --comment=<string>
234 An arbitrary comment. This option applies to job allocations.
235 --compress[=type]
238 Compress file before sending it to compute hosts. The optional
239 argument specifies the data compression library to be used.
240 Supported values are "lz4" (default) and "zlib". Some compres‐
241 sion libraries may be unavailable on some systems. For use with
242 the --bcast option. This option applies to step allocations.
243 -C, --constraint=<list>
246 Nodes can have features assigned to them by the Slurm adminis‐
247 trator. Users can specify which of these features are required
248 by their job using the constraint option. Only nodes having
249 features matching the job constraints will be used to satisfy
250 the request. Multiple constraints may be specified with AND,
251 OR, matching OR, resource counts, etc. (some operators are not
252 supported on all system types). Supported constraint options
253 include:
254 Single Name
256 Only nodes which have the specified feature will be used.
257 For example, --constraint="intel"
258 Node Count
260 A request can specify the number of nodes needed with
261 some feature by appending an asterisk and count after the
262 feature name. For example "--nodes=16 --con‐
263 straint=graphics*4 ..." indicates that the job requires
264 16 nodes and that at least four of those nodes must have
265 the feature "graphics."
266 AND If only nodes with all of specified features will be
268 used. The ampersand is used for an AND operator. For
269 example, --constraint="intel&gpu"
270 OR If only nodes with at least one of specified features
272 will be used. The vertical bar is used for an OR opera‐
273 tor. For example, --constraint="intel|amd"
274 Matching OR
276 If only one of a set of possible options should be used
277 for all allocated nodes, then use the OR operator and
278 enclose the options within square brackets. For example:
279 "--constraint=[rack1|rack2|rack3|rack4]" might be used to
280 specify that all nodes must be allocated on a single rack
281 of the cluster, but any of those four racks can be used.
282 Multiple Counts
284 Specific counts of multiple resources may be specified by
285 using the AND operator and enclosing the options within
286 square brackets. For example: "--con‐
287 straint=[rack1*2&rack2*4]" might be used to specify that
288 two nodes must be allocated from nodes with the feature
289 of "rack1" and four nodes must be allocated from nodes
290 with the feature "rack2".
291 NOTE: This construct does not support multiple Intel KNL
293 NUMA or MCDRAM modes. For example, while "--con‐
294 straint=[(knl&quad)*2&(knl&hemi)*4]" is not supported,
295 "--constraint=[haswell*2&(knl&hemi)*4]" is supported.
296 Specification of multiple KNL modes requires the use of a
297 heterogeneous job.
298 Parenthesis
301 Parenthesis can be used to group like node features
302 together. For example "--con‐
303 straint=[(knl&snc4&flat)*4&haswell*1]" might be used to
304 specify that four nodes with the features "knl", "snc4"
305 and "flat" plus one node with the feature "haswell" are
306 required. All options within parenthesis should be
307 grouped with AND (e.g. "&") operands.
308 WARNING: When srun is executed from within salloc or sbatch, the con‐
310 straint value can only contain a single feature name. None of the other
311 operators are currently supported for job steps.
312 This option applies to job and step allocations.
313 --contiguous
316 If set, then the allocated nodes must form a contiguous set.
317 Not honored with the topology/tree or topology/3d_torus plugins,
318 both of which can modify the node ordering. This option applies
319 to job allocations.
320 --cores-per-socket=<cores>
323 Restrict node selection to nodes with at least the specified
324 number of cores per socket. See additional information under -B
325 option above when task/affinity plugin is enabled. This option
326 applies to job allocations.
327 --cpu-bind=[{quiet,verbose},]type
330 Bind tasks to CPUs. Used only when the task/affinity or
331 task/cgroup plugin is enabled. NOTE: To have Slurm always
332 report on the selected CPU binding for all commands executed in
333 a shell, you can enable verbose mode by setting the
334 SLURM_CPU_BIND environment variable value to "verbose".
335 The following informational environment variables are set when
337 --cpu-bind is in use:
338 SLURM_CPU_BIND_VERBOSE
339 SLURM_CPU_BIND_TYPE
340 SLURM_CPU_BIND_LIST
341 See the ENVIRONMENT VARIABLES section for a more detailed
343 description of the individual SLURM_CPU_BIND variables. These
344 variable are available only if the task/affinity plugin is con‐
345 figured.
346 When using --cpus-per-task to run multithreaded tasks, be aware
348 that CPU binding is inherited from the parent of the process.
349 This means that the multithreaded task should either specify or
350 clear the CPU binding itself to avoid having all threads of the
351 multithreaded task use the same mask/CPU as the parent. Alter‐
352 natively, fat masks (masks which specify more than one allowed
353 CPU) could be used for the tasks in order to provide multiple
354 CPUs for the multithreaded tasks.
355 By default, a job step has access to every CPU allocated to the
357 job. To ensure that distinct CPUs are allocated to each job
358 step, use the --exclusive option.
359 Note that a job step can be allocated different numbers of CPUs
361 on each node or be allocated CPUs not starting at location zero.
362 Therefore one of the options which automatically generate the
363 task binding is recommended. Explicitly specified masks or
364 bindings are only honored when the job step has been allocated
365 every available CPU on the node.
366 Binding a task to a NUMA locality domain means to bind the task
368 to the set of CPUs that belong to the NUMA locality domain or
369 "NUMA node". If NUMA locality domain options are used on sys‐
370 tems with no NUMA support, then each socket is considered a
371 locality domain.
372 If the --cpu-bind option is not used, the default binding mode
374 will depend upon Slurm's configuration and the step's resource
375 allocation. If all allocated nodes have the same configured
376 CpuBind mode, that will be used. Otherwise if the job's Parti‐
377 tion has a configured CpuBind mode, that will be used. Other‐
378 wise if Slurm has a configured TaskPluginParam value, that mode
379 will be used. Otherwise automatic binding will be performed as
380 described below.
381 Auto Binding
384 Applies only when task/affinity is enabled. If the job
385 step allocation includes an allocation with a number of
386 sockets, cores, or threads equal to the number of tasks
387 times cpus-per-task, then the tasks will by default be
388 bound to the appropriate resources (auto binding). Dis‐
389 able this mode of operation by explicitly setting
390 "--cpu-bind=none". Use TaskPluginParam=auto‐
391 bind=[threads|cores|sockets] to set a default cpu binding
392 in case "auto binding" doesn't find a match.
393 Supported options include:
395 q[uiet]
397 Quietly bind before task runs (default)
398 v[erbose]
400 Verbosely report binding before task runs
401 no[ne] Do not bind tasks to CPUs (default unless auto
403 binding is applied)
404 rank Automatically bind by task rank. The lowest num‐
406 bered task on each node is bound to socket (or
407 core or thread) zero, etc. Not supported unless
408 the entire node is allocated to the job.
409 map_cpu:<list>
411 Bind by setting CPU masks on tasks (or ranks) as
412 specified where <list> is
413 <cpu_id_for_task_0>,<cpu_id_for_task_1>,... CPU
414 IDs are interpreted as decimal values unless they
415 are preceded with '0x' in which case they inter‐
416 preted as hexadecimal values. If the number of
417 tasks (or ranks) exceeds the number of elements in
418 this list, elements in the list will be reused as
419 needed starting from the beginning of the list.
420 To simplify support for large task counts, the
421 lists may follow a map with an asterisk and repe‐
422 tition count For example "map_cpu:0x0f*4,0xf0*4".
423 Not supported unless the entire node is allocated
424 to the job.
425 mask_cpu:<list>
427 Bind by setting CPU masks on tasks (or ranks) as
428 specified where <list> is
429 <cpu_mask_for_task_0>,<cpu_mask_for_task_1>,...
430 The mapping is specified for a node and identical
431 mapping is applied to the tasks on every node
432 (i.e. the lowest task ID on each node is mapped to
433 the first mask specified in the list, etc.). CPU
434 masks are always interpreted as hexadecimal values
435 but can be preceded with an optional '0x'. Not
436 supported unless the entire node is allocated to
437 the job. To simplify support for large task
438 counts, the lists may follow a map with an aster‐
439 isk and repetition count For example
440 "mask_cpu:0x0f*4,0xf0*4". Not supported unless
441 the entire node is allocated to the job.
442 rank_ldom
444 Bind to a NUMA locality domain by rank. Not sup‐
445 ported unless the entire node is allocated to the
446 job.
447 map_ldom:<list>
449 Bind by mapping NUMA locality domain IDs to tasks
450 as specified where <list> is
451 <ldom1>,<ldom2>,...<ldomN>. The locality domain
452 IDs are interpreted as decimal values unless they
453 are preceded with '0x' in which case they are
454 interpreted as hexadecimal values. Not supported
455 unless the entire node is allocated to the job.
456 mask_ldom:<list>
458 Bind by setting NUMA locality domain masks on
459 tasks as specified where <list> is
460 <mask1>,<mask2>,...<maskN>. NUMA locality domain
461 masks are always interpreted as hexadecimal values
462 but can be preceded with an optional '0x'. Not
463 supported unless the entire node is allocated to
464 the job.
465 sockets
467 Automatically generate masks binding tasks to
468 sockets. Only the CPUs on the socket which have
469 been allocated to the job will be used. If the
470 number of tasks differs from the number of allo‐
471 cated sockets this can result in sub-optimal bind‐
472 ing.
473 cores Automatically generate masks binding tasks to
475 cores. If the number of tasks differs from the
476 number of allocated cores this can result in
477 sub-optimal binding.
478 threads
480 Automatically generate masks binding tasks to
481 threads. If the number of tasks differs from the
482 number of allocated threads this can result in
483 sub-optimal binding.
484 ldoms Automatically generate masks binding tasks to NUMA
486 locality domains. If the number of tasks differs
487 from the number of allocated locality domains this
488 can result in sub-optimal binding.
489 boards Automatically generate masks binding tasks to
491 boards. If the number of tasks differs from the
492 number of allocated boards this can result in
493 sub-optimal binding. This option is supported by
494 the task/cgroup plugin only.
495 help Show help message for cpu-bind
497 This option applies to job and step allocations.
499 --cpu-freq =<p1[-p2[:p3]]>
502 Request that the job step initiated by this srun command be run
504 at some requested frequency if possible, on the CPUs selected
505 for the step on the compute node(s).
506 p1 can be [#### | low | medium | high | highm1] which will set
508 the frequency scaling_speed to the corresponding value, and set
509 the frequency scaling_governor to UserSpace. See below for defi‐
510 nition of the values.
511 p1 can be [Conservative | OnDemand | Performance | PowerSave]
513 which will set the scaling_governor to the corresponding value.
514 The governor has to be in the list set by the slurm.conf option
515 CpuFreqGovernors.
516 When p2 is present, p1 will be the minimum scaling frequency and
518 p2 will be the maximum scaling frequency.
519 p2 can be [#### | medium | high | highm1] p2 must be greater
521 than p1.
522 p3 can be [Conservative | OnDemand | Performance | PowerSave |
524 UserSpace] which will set the governor to the corresponding
525 value.
526 If p3 is UserSpace, the frequency scaling_speed will be set by a
528 power or energy aware scheduling strategy to a value between p1
529 and p2 that lets the job run within the site's power goal. The
530 job may be delayed if p1 is higher than a frequency that allows
531 the job to run within the goal.
532 If the current frequency is < min, it will be set to min. Like‐
534 wise, if the current frequency is > max, it will be set to max.
535 Acceptable values at present include:
537 #### frequency in kilohertz
539 Low the lowest available frequency
541 High the highest available frequency
543 HighM1 (high minus one) will select the next highest
545 available frequency
546 Medium attempts to set a frequency in the middle of the
548 available range
549 Conservative attempts to use the Conservative CPU governor
551 OnDemand attempts to use the OnDemand CPU governor (the
553 default value)
554 Performance attempts to use the Performance CPU governor
556 PowerSave attempts to use the PowerSave CPU governor
558 UserSpace attempts to use the UserSpace CPU governor
560 The following informational environment variable is set
563 in the job
564 step when --cpu-freq option is requested.
565 SLURM_CPU_FREQ_REQ
566 This environment variable can also be used to supply the value
568 for the CPU frequency request if it is set when the 'srun' com‐
569 mand is issued. The --cpu-freq on the command line will over‐
570 ride the environment variable value. The form on the environ‐
571 ment variable is the same as the command line. See the ENVIRON‐
572 MENT VARIABLES section for a description of the
573 SLURM_CPU_FREQ_REQ variable.
574 NOTE: This parameter is treated as a request, not a requirement.
576 If the job step's node does not support setting the CPU fre‐
577 quency, or the requested value is outside the bounds of the
578 legal frequencies, an error is logged, but the job step is
579 allowed to continue.
580 NOTE: Setting the frequency for just the CPUs of the job step
582 implies that the tasks are confined to those CPUs. If task con‐
583 finement (i.e., TaskPlugin=task/affinity or TaskPlu‐
584 gin=task/cgroup with the "ConstrainCores" option) is not config‐
585 ured, this parameter is ignored.
586 NOTE: When the step completes, the frequency and governor of
588 each selected CPU is reset to the previous values.
589 NOTE: When submitting jobs with the --cpu-freq option with lin‐
591 uxproc as the ProctrackType can cause jobs to run too quickly
592 before Accounting is able to poll for job information. As a
593 result not all of accounting information will be present.
594 This option applies to job and step allocations.
596 --cpus-per-gpu=<ncpus>
599 Advise Slurm that ensuing job steps will require ncpus proces‐
600 sors per allocated GPU. Requires the --gpus option. Not com‐
601 patible with the --cpus-per-task option.
602 -c, --cpus-per-task=<ncpus>
605 Request that ncpus be allocated per process. This may be useful
606 if the job is multithreaded and requires more than one CPU per
607 task for optimal performance. The default is one CPU per
608 process. If -c is specified without -n, as many tasks will be
609 allocated per node as possible while satisfying the -c restric‐
610 tion. For instance on a cluster with 8 CPUs per node, a job
611 request for 4 nodes and 3 CPUs per task may be allocated 3 or 6
612 CPUs per node (1 or 2 tasks per node) depending upon resource
613 consumption by other jobs. Such a job may be unable to execute
614 more than a total of 4 tasks. This option may also be useful to
615 spawn tasks without allocating resources to the job step from
616 the job's allocation when running multiple job steps with the
617 --exclusive option.
618 WARNING: There are configurations and options interpreted dif‐
620 ferently by job and job step requests which can result in incon‐
621 sistencies for this option. For example srun -c2
622 --threads-per-core=1 prog may allocate two cores for the job,
623 but if each of those cores contains two threads, the job alloca‐
624 tion will include four CPUs. The job step allocation will then
625 launch two threads per CPU for a total of two tasks.
626 WARNING: When srun is executed from within salloc or sbatch,
628 there are configurations and options which can result in incon‐
629 sistent allocations when -c has a value greater than -c on sal‐
630 loc or sbatch.
631 This option applies to job allocations.
633 --deadline=<OPT>
636 remove the job if no ending is possible before this deadline
637 (start > (deadline - time[-min])). Default is no deadline.
638 Valid time formats are:
639 HH:MM[:SS] [AM|PM]
640 MMDD[YY] or MM/DD[/YY] or MM.DD[.YY]
641 MM/DD[/YY]-HH:MM[:SS]
642 YYYY-MM-DD[THH:MM[:SS]]]
643 This option applies only to job allocations.
645 --delay-boot=<minutes>
648 Do not reboot nodes in order to satisfied this job's feature
649 specification if the job has been eligible to run for less than
650 this time period. If the job has waited for less than the spec‐
651 ified period, it will use only nodes which already have the
652 specified features. The argument is in units of minutes. A
653 default value may be set by a system administrator using the
654 delay_boot option of the SchedulerParameters configuration
655 parameter in the slurm.conf file, otherwise the default value is
656 zero (no delay).
657 This option applies only to job allocations.
659 -d, --dependency=<dependency_list>
662 Defer the start of this job until the specified dependencies
663 have been satisfied completed. This option does not apply to job
664 steps (executions of srun within an existing salloc or sbatch
665 allocation) only to job allocations. <dependency_list> is of
666 the form <type:job_id[:job_id][,type:job_id[:job_id]]> or
667 <type:job_id[:job_id][?type:job_id[:job_id]]>. All dependencies
668 must be satisfied if the "," separator is used. Any dependency
669 may be satisfied if the "?" separator is used. Many jobs can
670 share the same dependency and these jobs may even belong to dif‐
671 ferent users. The value may be changed after job submission
672 using the scontrol command. Once a job dependency fails due to
673 the termination state of a preceding job, the dependent job will
674 never be run, even if the preceding job is requeued and has a
675 different termination state in a subsequent execution. This
676 option applies to job allocations.
677 after:job_id[:jobid...]
679 This job can begin execution after the specified jobs
680 have begun execution.
681 afterany:job_id[:jobid...]
683 This job can begin execution after the specified jobs
684 have terminated.
685 afterburstbuffer:job_id[:jobid...]
687 This job can begin execution after the specified jobs
688 have terminated and any associated burst buffer stage out
689 operations have completed.
690 aftercorr:job_id[:jobid...]
692 A task of this job array can begin execution after the
693 corresponding task ID in the specified job has completed
694 successfully (ran to completion with an exit code of
695 zero).
696 afternotok:job_id[:jobid...]
698 This job can begin execution after the specified jobs
699 have terminated in some failed state (non-zero exit code,
700 node failure, timed out, etc).
701 afterok:job_id[:jobid...]
703 This job can begin execution after the specified jobs
704 have successfully executed (ran to completion with an
705 exit code of zero).
706 expand:job_id
708 Resources allocated to this job should be used to expand
709 the specified job. The job to expand must share the same
710 QOS (Quality of Service) and partition. Gang scheduling
711 of resources in the partition is also not supported.
712 singleton
714 This job can begin execution after any previously
715 launched jobs sharing the same job name and user have
716 terminated. In other words, only one job by that name
717 and owned by that user can be running or suspended at any
718 point in time.
719 -D, --chdir=<path>
722 Have the remote processes do a chdir to path before beginning
723 execution. The default is to chdir to the current working direc‐
724 tory of the srun process. The path can be specified as full path
725 or relative path to the directory where the command is executed.
726 This option applies to job allocations.
727 -e, --error=<filename pattern>
730 Specify how stderr is to be redirected. By default in interac‐
731 tive mode, srun redirects stderr to the same file as stdout, if
732 one is specified. The --error option is provided to allow stdout
733 and stderr to be redirected to different locations. See IO Re‐
734 direction below for more options. If the specified file already
735 exists, it will be overwritten. This option applies to job and
736 step allocations.
737 -E, --preserve-env
740 Pass the current values of environment variables SLURM_JOB_NODES
741 and SLURM_NTASKS through to the executable, rather than comput‐
742 ing them from commandline parameters. This option applies to job
743 allocations.
744 --epilog=<executable>
747 srun will run executable just after the job step completes. The
748 command line arguments for executable will be the command and
749 arguments of the job step. If executable is "none", then no
750 srun epilog will be run. This parameter overrides the SrunEpilog
751 parameter in slurm.conf. This parameter is completely indepen‐
752 dent from the Epilog parameter in slurm.conf. This option
753 applies to job allocations.
754 --exclusive[=user|mcs]
758 This option applies to job and job step allocations, and has two
759 slightly different meanings for each one. When used to initiate
760 a job, the job allocation cannot share nodes with other running
761 jobs (or just other users with the "=user" option or "=mcs"
762 option). The default shared/exclusive behavior depends on sys‐
763 tem configuration and the partition's OverSubscribe option takes
764 precedence over the job's option.
765 This option can also be used when initiating more than one job
767 step within an existing resource allocation, where you want sep‐
768 arate processors to be dedicated to each job step. If sufficient
769 processors are not available to initiate the job step, it will
770 be deferred. This can be thought of as providing a mechanism for
771 resource management to the job within it's allocation.
772 The exclusive allocation of CPUs only applies to job steps
774 explicitly invoked with the --exclusive option. For example, a
775 job might be allocated one node with four CPUs and a remote
776 shell invoked on the allocated node. If that shell is not
777 invoked with the --exclusive option, then it may create a job
778 step with four tasks using the --exclusive option and not con‐
779 flict with the remote shell's resource allocation. Use the
780 --exclusive option to invoke every job step to ensure distinct
781 resources for each step.
782 Note that all CPUs allocated to a job are available to each job
784 step unless the --exclusive option is used plus task affinity is
785 configured. Since resource management is provided by processor,
786 the --ntasks option must be specified, but the following options
787 should NOT be specified --relative, --distribution=arbitrary.
788 See EXAMPLE below.
789 --export=<environment variables [ALL] | NONE>
792 Identify which environment variables are propagated to the
793 launched application. By default, all are propagated. Multiple
794 environment variable names should be comma separated. Environ‐
795 ment variable names may be specified to propagate the current
796 value (e.g. "--export=EDITOR") or specific values may be
797 exported (e.g. "--export=EDITOR=/bin/emacs"). In these two exam‐
798 ples, the propagated environment will only contain the variable
799 EDITOR. If one desires to add to the environment instead of
800 replacing it, have the argument include ALL (e.g.
801 "--export=ALL,EDITOR=/bin/emacs"). This will propagate EDITOR
802 along with the current environment. Unlike sbatch, if ALL is
803 specified, any additional specified environment variables are
804 ignored. If one desires no environment variables be propagated,
805 use the argument NONE. Regardless of this setting, the appro‐
806 priate SLURM_* task environment variables are always exported to
807 the environment. srun may deviate from the above behavior if
808 the default launch plugin, launch/slurm, is not used.
809 -F, --nodefile=<node file>
812 Much like --nodelist, but the list is contained in a file of
813 name node file. The node names of the list may also span multi‐
814 ple lines in the file. Duplicate node names in the file will
815 be ignored. The order of the node names in the list is not
816 important; the node names will be sorted by Slurm.
817 --gid=<group>
820 If srun is run as root, and the --gid option is used, submit the
821 job with group's group access permissions. group may be the
822 group name or the numerical group ID. This option applies to job
823 allocations.
824 -G, --gpus=[<type>:]<number>
827 Specify the total number of GPUs required for the job. An
828 optional GPU type specification can be supplied. For example
829 "--gpus=volta:3". Multiple options can be requested in a comma
830 separated list, for example: "--gpus=volta:3,kepler:1". See
831 also the --gpus-per-node, --gpus-per-socket and --gpus-per-task
832 options.
833 --gpu-bind=<type>
836 Bind tasks to specific GPUs. By default every spawned task can
837 access every GPU allocated to the job.
838 Supported type options:
840 closest Bind each task to the GPU(s) which are closest. In a
842 NUMA environment, each task may be bound to more than
843 one GPU (i.e. all GPUs in that NUMA environment).
844 map_gpu:<list>
846 Bind by setting GPU masks on tasks (or ranks) as spec‐
847 ified where <list> is
848 <gpu_id_for_task_0>,<gpu_id_for_task_1>,... GPU IDs
849 are interpreted as decimal values unless they are pre‐
850 ceded with '0x' in which case they interpreted as
851 hexadecimal values. If the number of tasks (or ranks)
852 exceeds the number of elements in this list, elements
853 in the list will be reused as needed starting from the
854 beginning of the list. To simplify support for large
855 task counts, the lists may follow a map with an aster‐
856 isk and repetition count. For example
857 "map_cpu:0*4,1*4". Not supported unless the entire
858 node is allocated to the job.
859 mask_gpu:<list>
861 Bind by setting GPU masks on tasks (or ranks) as spec‐
862 ified where <list> is
863 <gpu_mask_for_task_0>,<gpu_mask_for_task_1>,... The
864 mapping is specified for a node and identical mapping
865 is applied to the tasks on every node (i.e. the lowest
866 task ID on each node is mapped to the first mask spec‐
867 ified in the list, etc.). GPU masks are always inter‐
868 preted as hexadecimal values but can be preceded with
869 an optional '0x'. Not supported unless the entire node
870 is allocated to the job. To simplify support for large
871 task counts, the lists may follow a map with an aster‐
872 isk and repetition count. For example
873 "mask_gpu:0x0f*4,0xf0*4". Not supported unless the
874 entire node is allocated to the job.
875 --gpu-freq=[<type]=value>[,<type=value>][,verbose]
878 Request that GPUs allocated to the job are configured with spe‐
879 cific frequency values. This option can be used to indepen‐
880 dently configure the GPU and its memory frequencies. After the
881 job is completed, the frequencies of all affected GPUs will be
882 reset to the highest possible values. In some cases, system
883 power caps may override the requested values. The field type
884 can be "memory". If type is not specified, the GPU frequency is
885 implied. The value field can either be "low", "medium", "high",
886 "highm1" or a numeric value in megahertz (MHz). If the speci‐
887 fied numeric value is not possible, a value as close as possible
888 will be used. See below for definition of the values. The ver‐
889 bose option causes current GPU frequency information to be
890 logged. Examples of use include "--gpu-freq=medium,memory=high"
891 and "--gpu-freq=450".
892 Supported value definitions:
894 low the lowest available frequency.
896 medium attempts to set a frequency in the middle of the
898 available range.
899 high the highest available frequency.
901 highm1 (high minus one) will select the next highest avail‐
903 able frequency.
904 --gpus-per-node=[<type>:]<number>
907 Specify the number of GPUs required for the job on each node
908 included in the job's resource allocation. An optional GPU type
909 specification can be supplied. For example
910 "--gpus-per-node=volta:3". Multiple options can be requested in
911 a comma separated list, for example:
912 "--gpus-per-node=volta:3,kepler:1". See also the --gpus,
913 --gpus-per-socket and --gpus-per-task options.
914 --gpus-per-socket=[<type>:]<number>
917 Specify the number of GPUs required for the job on each socket
918 included in the job's resource allocation. An optional GPU type
919 specification can be supplied. For example
920 "--gpus-per-socket=volta:3". Multiple options can be requested
921 in a comma separated list, for example:
922 "--gpus-per-socket=volta:3,kepler:1". Requires job to specify a
923 sockets per node count ( --sockets-per-node). See also the
924 --gpus, --gpus-per-node and --gpus-per-task options. This
925 option applies to job allocations.
926 --gpus-per-task=[<type>:]<number>
929 Specify the number of GPUs required for the job on each task to
930 be spawned in the job's resource allocation. An optional GPU
931 type specification can be supplied. This option requires the
932 specification of a task count. For example
933 "--gpus-per-task=volta:1". Multiple options can be requested in
934 a comma separated list, for example:
935 "--gpus-per-task=volta:3,kepler:1". Requires job to specify a
936 task count (--nodes). See also the --gpus, --gpus-per-socket
937 and --gpus-per-node options.
938 --gres=<list>
941 Specifies a comma delimited list of generic consumable
942 resources. The format of each entry on the list is
943 "name[[:type]:count]". The name is that of the consumable
944 resource. The count is the number of those resources with a
945 default value of 1. The count can have a suffix of "k" or "K"
946 (multiple of 1024), "m" or "M" (multiple of 1024 x 1024), "g" or
947 "G" (multiple of 1024 x 1024 x 1024), "t" or "T" (multiple of
948 1024 x 1024 x 1024 x 1024), "p" or "P" (multiple of 1024 x 1024
949 x 1024 x 1024 x 1024). The specified resources will be allo‐
950 cated to the job on each node. The available generic consumable
951 resources is configurable by the system administrator. A list
952 of available generic consumable resources will be printed and
953 the command will exit if the option argument is "help". Exam‐
954 ples of use include "--gres=gpu:2,mic:1", "--gres=gpu:kepler:2",
955 and "--gres=help". NOTE: This option applies to job and step
956 allocations. By default, a job step is allocated all of the
957 generic resources that have allocated to the job. To change the
958 behavior so that each job step is allocated no generic
959 resources, explicitly set the value of --gres to specify zero
960 counts for each generic resource OR set "--gres=none" OR set the
961 SLURM_STEP_GRES environment variable to "none".
962 --gres-flags=<type>
965 Specify generic resource task binding options. This option
966 applies to job allocations.
967 disable-binding
969 Disable filtering of CPUs with respect to generic
970 resource locality. This option is currently required to
971 use more CPUs than are bound to a GRES (i.e. if a GPU is
972 bound to the CPUs on one socket, but resources on more
973 than one socket are required to run the job). This
974 option may permit a job to be allocated resources sooner
975 than otherwise possible, but may result in lower job per‐
976 formance.
977 enforce-binding
979 The only CPUs available to the job will be those bound to
980 the selected GRES (i.e. the CPUs identified in the
981 gres.conf file will be strictly enforced). This option
982 may result in delayed initiation of a job. For example a
983 job requiring two GPUs and one CPU will be delayed until
984 both GPUs on a single socket are available rather than
985 using GPUs bound to separate sockets, however the appli‐
986 cation performance may be improved due to improved commu‐
987 nication speed. Requires the node to be configured with
988 more than one socket and resource filtering will be per‐
989 formed on a per-socket basis.
990 -H, --hold
993 Specify the job is to be submitted in a held state (priority of
994 zero). A held job can now be released using scontrol to reset
995 its priority (e.g. "scontrol release <job_id>"). This option
996 applies to job allocations.
997 -h, --help
1000 Display help information and exit.
1001 --hint=<type>
1004 Bind tasks according to application hints.
1005 compute_bound
1007 Select settings for compute bound applications: use all
1008 cores in each socket, one thread per core.
1009 memory_bound
1011 Select settings for memory bound applications: use only
1012 one core in each socket, one thread per core.
1013 [no]multithread
1015 [don't] use extra threads with in-core multi-threading
1016 which can benefit communication intensive applications.
1017 Only supported with the task/affinity plugin.
1018 help show this help message
1020 This option applies to job allocations.
1022 -I, --immediate[=<seconds>]
1025 exit if resources are not available within the time period spec‐
1026 ified. If no argument is given, resources must be available
1027 immediately for the request to succeed. By default, --immediate
1028 is off, and the command will block until resources become avail‐
1029 able. Since this option's argument is optional, for proper pars‐
1030 ing the single letter option must be followed immediately with
1031 the value and not include a space between them. For example
1032 "-I60" and not "-I 60". This option applies to job and step
1033 allocations.
1034 -i, --input=<mode>
1037 Specify how stdin is to redirected. By default, srun redirects
1038 stdin from the terminal all tasks. See IO Redirection below for
1039 more options. For OS X, the poll() function does not support
1040 stdin, so input from a terminal is not possible. This option
1041 applies to job and step allocations.
1042 -J, --job-name=<jobname>
1045 Specify a name for the job. The specified name will appear along
1046 with the job id number when querying running jobs on the system.
1047 The default is the supplied executable program's name. NOTE:
1048 This information may be written to the slurm_jobacct.log file.
1049 This file is space delimited so if a space is used in the job‐
1050 name name it will cause problems in properly displaying the con‐
1051 tents of the slurm_jobacct.log file when the sacct command is
1052 used. This option applies to job and step allocations.
1053 --jobid=<jobid>
1056 Initiate a job step under an already allocated job with job id
1057 id. Using this option will cause srun to behave exactly as if
1058 the SLURM_JOB_ID environment variable was set. This option
1059 applies to step allocations.
1060 -K, --kill-on-bad-exit[=0|1]
1063 Controls whether or not to terminate a step if any task exits
1064 with a non-zero exit code. If this option is not specified, the
1065 default action will be based upon the Slurm configuration param‐
1066 eter of KillOnBadExit. If this option is specified, it will take
1067 precedence over KillOnBadExit. An option argument of zero will
1068 not terminate the job. A non-zero argument or no argument will
1069 terminate the job. Note: This option takes precedence over the
1070 -W, --wait option to terminate the job immediately if a task
1071 exits with a non-zero exit code. Since this option's argument
1072 is optional, for proper parsing the single letter option must be
1073 followed immediately with the value and not include a space
1074 between them. For example "-K1" and not "-K 1".
1075 -k, --no-kill [=off]
1078 Do not automatically terminate a job if one of the nodes it has
1079 been allocated fails. This option applies to job and step allo‐
1080 cations. The job will assume all responsibilities for
1081 fault-tolerance. Tasks launch using this option will not be
1082 considered terminated (e.g. -K, --kill-on-bad-exit and -W,
1083 --wait options will have no effect upon the job step). The
1084 active job step (MPI job) will likely suffer a fatal error, but
1085 subsequent job steps may be run if this option is specified.
1086 Specify an optional argument of "off" disable the effect of the
1088 SLURM_NO_KILL environment variable.
1089 The default action is to terminate the job upon node failure.
1091 -l, --label
1094 Prepend task number to lines of stdout/err. The --label option
1095 will prepend lines of output with the remote task id. This
1096 option applies to step allocations.
1097 -L, --licenses=<license>
1100 Specification of licenses (or other resources available on all
1101 nodes of the cluster) which must be allocated to this job.
1102 License names can be followed by a colon and count (the default
1103 count is one). Multiple license names should be comma separated
1104 (e.g. "--licenses=foo:4,bar"). This option applies to job allo‐
1105 cations.
1106 -M, --clusters=<string>
1109 Clusters to issue commands to. Multiple cluster names may be
1110 comma separated. The job will be submitted to the one cluster
1111 providing the earliest expected job initiation time. The default
1112 value is the current cluster. A value of 'all' will query to run
1113 on all clusters. Note the --export option to control environ‐
1114 ment variables exported between clusters. This option applies
1115 only to job allocations. Note that the SlurmDBD must be up for
1116 this option to work properly.
1117 -m, --distribution=
1120 *|block|cyclic|arbitrary|plane=<options>
1121 [:*|block|cyclic|fcyclic[:*|block|
1122 cyclic|fcyclic]][,Pack|NoPack]
1123 Specify alternate distribution methods for remote processes.
1125 This option controls the distribution of tasks to the nodes on
1126 which resources have been allocated, and the distribution of
1127 those resources to tasks for binding (task affinity). The first
1128 distribution method (before the first ":") controls the distri‐
1129 bution of tasks to nodes. The second distribution method (after
1130 the first ":") controls the distribution of allocated CPUs
1131 across sockets for binding to tasks. The third distribution
1132 method (after the second ":") controls the distribution of allo‐
1133 cated CPUs across cores for binding to tasks. The second and
1134 third distributions apply only if task affinity is enabled. The
1135 third distribution is supported only if the task/cgroup plugin
1136 is configured. The default value for each distribution type is
1137 specified by *.
1138 Note that with select/cons_res, the number of CPUs allocated on
1140 each socket and node may be different. Refer to
1141 https://slurm.schedmd.com/mc_support.html for more information
1142 on resource allocation, distribution of tasks to nodes, and
1143 binding of tasks to CPUs.
1144 First distribution method (distribution of tasks across nodes):
1145 * Use the default method for distributing tasks to nodes
1148 (block).
1149 block The block distribution method will distribute tasks to a
1151 node such that consecutive tasks share a node. For exam‐
1152 ple, consider an allocation of three nodes each with two
1153 cpus. A four-task block distribution request will dis‐
1154 tribute those tasks to the nodes with tasks one and two
1155 on the first node, task three on the second node, and
1156 task four on the third node. Block distribution is the
1157 default behavior if the number of tasks exceeds the num‐
1158 ber of allocated nodes.
1159 cyclic The cyclic distribution method will distribute tasks to a
1161 node such that consecutive tasks are distributed over
1162 consecutive nodes (in a round-robin fashion). For exam‐
1163 ple, consider an allocation of three nodes each with two
1164 cpus. A four-task cyclic distribution request will dis‐
1165 tribute those tasks to the nodes with tasks one and four
1166 on the first node, task two on the second node, and task
1167 three on the third node. Note that when SelectType is
1168 select/cons_res, the same number of CPUs may not be allo‐
1169 cated on each node. Task distribution will be round-robin
1170 among all the nodes with CPUs yet to be assigned to
1171 tasks. Cyclic distribution is the default behavior if
1172 the number of tasks is no larger than the number of allo‐
1173 cated nodes.
1174 plane The tasks are distributed in blocks of a specified size.
1176 The options include a number representing the size of the
1177 task block. This is followed by an optional specifica‐
1178 tion of the task distribution scheme within a block of
1179 tasks and between the blocks of tasks. The number of
1180 tasks distributed to each node is the same as for cyclic
1181 distribution, but the taskids assigned to each node
1182 depend on the plane size. For more details (including
1183 examples and diagrams), please see
1184 https://slurm.schedmd.com/mc_support.html
1185 and
1186 https://slurm.schedmd.com/dist_plane.html
1187 arbitrary
1189 The arbitrary method of distribution will allocate pro‐
1190 cesses in-order as listed in file designated by the envi‐
1191 ronment variable SLURM_HOSTFILE. If this variable is
1192 listed it will over ride any other method specified. If
1193 not set the method will default to block. Inside the
1194 hostfile must contain at minimum the number of hosts
1195 requested and be one per line or comma separated. If
1196 specifying a task count (-n, --ntasks=<number>), your
1197 tasks will be laid out on the nodes in the order of the
1198 file.
1199 NOTE: The arbitrary distribution option on a job alloca‐
1200 tion only controls the nodes to be allocated to the job
1201 and not the allocation of CPUs on those nodes. This
1202 option is meant primarily to control a job step's task
1203 layout in an existing job allocation for the srun com‐
1204 mand.
1205 NOTE: If number of tasks is given and a list of requested
1206 nodes is also given the number of nodes used from that
1207 list will be reduced to match that of the number of tasks
1208 if the number of nodes in the list is greater than the
1209 number of tasks.
1210 Second distribution method (distribution of CPUs across sockets
1213 for binding):
1214 * Use the default method for distributing CPUs across sock‐
1217 ets (cyclic).
1218 block The block distribution method will distribute allocated
1220 CPUs consecutively from the same socket for binding to
1221 tasks, before using the next consecutive socket.
1222 cyclic The cyclic distribution method will distribute allocated
1224 CPUs for binding to a given task consecutively from the
1225 same socket, and from the next consecutive socket for the
1226 next task, in a round-robin fashion across sockets.
1227 fcyclic
1229 The fcyclic distribution method will distribute allocated
1230 CPUs for binding to tasks from consecutive sockets in a
1231 round-robin fashion across the sockets.
1232 Third distribution method (distribution of CPUs across cores for
1235 binding):
1236 * Use the default method for distributing CPUs across cores
1239 (inherited from second distribution method).
1240 block The block distribution method will distribute allocated
1242 CPUs consecutively from the same core for binding to
1243 tasks, before using the next consecutive core.
1244 cyclic The cyclic distribution method will distribute allocated
1246 CPUs for binding to a given task consecutively from the
1247 same core, and from the next consecutive core for the
1248 next task, in a round-robin fashion across cores.
1249 fcyclic
1251 The fcyclic distribution method will distribute allocated
1252 CPUs for binding to tasks from consecutive cores in a
1253 round-robin fashion across the cores.
1254 Optional control for task distribution over nodes:
1258 Pack Rather than evenly distributing a job step's tasks evenly
1261 across it's allocated nodes, pack them as tightly as pos‐
1262 sible on the nodes.
1263 NoPack Rather than packing a job step's tasks as tightly as pos‐
1265 sible on the nodes, distribute them evenly. This user
1266 option will supersede the SelectTypeParameters
1267 CR_Pack_Nodes configuration parameter.
1268 This option applies to job and step allocations.
1270 --mail-type=<type>
1273 Notify user by email when certain event types occur. Valid type
1274 values are NONE, BEGIN, END, FAIL, REQUEUE, ALL (equivalent to
1275 BEGIN, END, FAIL, REQUEUE, and STAGE_OUT), STAGE_OUT (burst buf‐
1276 fer stage out and teardown completed), TIME_LIMIT, TIME_LIMIT_90
1277 (reached 90 percent of time limit), TIME_LIMIT_80 (reached 80
1278 percent of time limit), and TIME_LIMIT_50 (reached 50 percent of
1279 time limit). Multiple type values may be specified in a comma
1280 separated list. The user to be notified is indicated with
1281 --mail-user. This option applies to job allocations.
1282 --mail-user=<user>
1285 User to receive email notification of state changes as defined
1286 by --mail-type. The default value is the submitting user. This
1287 option applies to job allocations.
1288 --mcs-label=<mcs>
1291 Used only when the mcs/group plugin is enabled. This parameter
1292 is a group among the groups of the user. Default value is cal‐
1293 culated by the Plugin mcs if it's enabled. This option applies
1294 to job allocations.
1295 --mem=<size[units]>
1298 Specify the real memory required per node. Default units are
1299 megabytes unless the SchedulerParameters configuration parameter
1300 includes the "default_gbytes" option for gigabytes. Different
1301 units can be specified using the suffix [K|M|G|T]. Default
1302 value is DefMemPerNode and the maximum value is MaxMemPerNode.
1303 If configured, both of parameters can be seen using the scontrol
1304 show config command. This parameter would generally be used if
1305 whole nodes are allocated to jobs (SelectType=select/linear).
1306 Specifying a memory limit of zero for a job step will restrict
1307 the job step to the amount of memory allocated to the job, but
1308 not remove any of the job's memory allocation from being avail‐
1309 able to other job steps. Also see --mem-per-cpu and
1310 --mem-per-gpu. The --mem, --mem-per-cpu and --mem-per-gpu
1311 options are mutually exclusive. If --mem, --mem-per-cpu or
1312 --mem-per-gpu are specified as command line arguments, then they
1313 will take precedence over the environment (potentially inherited
1314 from salloc or sbatch).
1315 NOTE: A memory size specification of zero is treated as a spe‐
1317 cial case and grants the job access to all of the memory on each
1318 node for newly submitted jobs and all available job memory to a
1319 new job steps.
1320 Specifying new memory limits for job steps are only advisory.
1322 If the job is allocated multiple nodes in a heterogeneous clus‐
1324 ter, the memory limit on each node will be that of the node in
1325 the allocation with the smallest memory size (same limit will
1326 apply to every node in the job's allocation).
1327 NOTE: Enforcement of memory limits currently relies upon the
1329 task/cgroup plugin or enabling of accounting, which samples mem‐
1330 ory use on a periodic basis (data need not be stored, just col‐
1331 lected). In both cases memory use is based upon the job's Resi‐
1332 dent Set Size (RSS). A task may exceed the memory limit until
1333 the next periodic accounting sample.
1334 This option applies to job and step allocations.
1336 --mem-per-cpu=<size[units]>
1339 Minimum memory required per allocated CPU. Default units are
1340 megabytes unless the SchedulerParameters configuration parameter
1341 includes the "default_gbytes" option for gigabytes. Different
1342 units can be specified using the suffix [K|M|G|T]. Default
1343 value is DefMemPerCPU and the maximum value is MaxMemPerCPU (see
1344 exception below). If configured, both of parameters can be seen
1345 using the scontrol show config command. Note that if the job's
1346 --mem-per-cpu value exceeds the configured MaxMemPerCPU, then
1347 the user's limit will be treated as a memory limit per task;
1348 --mem-per-cpu will be reduced to a value no larger than MaxMem‐
1349 PerCPU; --cpus-per-task will be set and the value of
1350 --cpus-per-task multiplied by the new --mem-per-cpu value will
1351 equal the original --mem-per-cpu value specified by the user.
1352 This parameter would generally be used if individual processors
1353 are allocated to jobs (SelectType=select/cons_res). If
1354 resources are allocated by the core, socket or whole nodes; the
1355 number of CPUs allocated to a job may be higher than the task
1356 count and the value of --mem-per-cpu should be adjusted accord‐
1357 ingly. Specifying a memory limit of zero for a job step will
1358 restrict the job step to the amount of memory allocated to the
1359 job, but not remove any of the job's memory allocation from
1360 being available to other job steps. Also see --mem and
1361 --mem-per-gpu. The --mem, --mem-per-cpu and --mem-per-gpu
1362 options are mutually exclusive.
1363 NOTE:If the final amount of memory requested by job (eg.: when
1365 --mem-per-cpu use with --exclusive option) can't be satisfied by
1366 any of nodes configured in the partition, the job will be
1367 rejected.
1368 --mem-per-gpu=<size[units]>
1371 Minimum memory required per allocated GPU. Default units are
1372 megabytes unless the SchedulerParameters configuration parameter
1373 includes the "default_gbytes" option for gigabytes. Different
1374 units can be specified using the suffix [K|M|G|T]. Default
1375 value is DefMemPerGPU and is available on both a global and per
1376 partition basis. If configured, the parameters can be seen
1377 using the scontrol show config and scontrol show partition com‐
1378 mands. Also see --mem. The --mem, --mem-per-cpu and
1379 --mem-per-gpu options are mutually exclusive.
1380 --mem-bind=[{quiet,verbose},]type
1383 Bind tasks to memory. Used only when the task/affinity plugin is
1384 enabled and the NUMA memory functions are available. Note that
1385 the resolution of CPU and memory binding may differ on some
1386 architectures. For example, CPU binding may be performed at the
1387 level of the cores within a processor while memory binding will
1388 be performed at the level of nodes, where the definition of
1389 "nodes" may differ from system to system. By default no memory
1390 binding is performed; any task using any CPU can use any memory.
1391 This option is typically used to ensure that each task is bound
1392 to the memory closest to it's assigned CPU. The use of any type
1393 other than "none" or "local" is not recommended. If you want
1394 greater control, try running a simple test code with the options
1395 "--cpu-bind=verbose,none --mem-bind=verbose,none" to determine
1396 the specific configuration.
1397 NOTE: To have Slurm always report on the selected memory binding
1399 for all commands executed in a shell, you can enable verbose
1400 mode by setting the SLURM_MEM_BIND environment variable value to
1401 "verbose".
1402 The following informational environment variables are set when
1404 --mem-bind is in use:
1405 SLURM_MEM_BIND_LIST
1407 SLURM_MEM_BIND_PREFER
1408 SLURM_MEM_BIND_SORT
1409 SLURM_MEM_BIND_TYPE
1410 SLURM_MEM_BIND_VERBOSE
1411 See the ENVIRONMENT VARIABLES section for a more detailed
1413 description of the individual SLURM_MEM_BIND* variables.
1414 Supported options include:
1416 help show this help message
1418 local Use memory local to the processor in use
1420 map_mem:<list>
1422 Bind by setting memory masks on tasks (or ranks) as spec‐
1423 ified where <list> is
1424 <numa_id_for_task_0>,<numa_id_for_task_1>,... The map‐
1425 ping is specified for a node and identical mapping is
1426 applied to the tasks on every node (i.e. the lowest task
1427 ID on each node is mapped to the first ID specified in
1428 the list, etc.). NUMA IDs are interpreted as decimal
1429 values unless they are preceded with '0x' in which case
1430 they interpreted as hexadecimal values. If the number of
1431 tasks (or ranks) exceeds the number of elements in this
1432 list, elements in the list will be reused as needed
1433 starting from the beginning of the list. To simplify
1434 support for large task counts, the lists may follow a map
1435 with an asterisk and repetition count For example
1436 "map_mem:0x0f*4,0xf0*4". Not supported unless the entire
1437 node is allocated to the job.
1438 mask_mem:<list>
1440 Bind by setting memory masks on tasks (or ranks) as spec‐
1441 ified where <list> is
1442 <numa_mask_for_task_0>,<numa_mask_for_task_1>,... The
1443 mapping is specified for a node and identical mapping is
1444 applied to the tasks on every node (i.e. the lowest task
1445 ID on each node is mapped to the first mask specified in
1446 the list, etc.). NUMA masks are always interpreted as
1447 hexadecimal values. Note that masks must be preceded
1448 with a '0x' if they don't begin with [0-9] so they are
1449 seen as numerical values. If the number of tasks (or
1450 ranks) exceeds the number of elements in this list, ele‐
1451 ments in the list will be reused as needed starting from
1452 the beginning of the list. To simplify support for large
1453 task counts, the lists may follow a mask with an asterisk
1454 and repetition count For example "mask_mem:0*4,1*4". Not
1455 supported unless the entire node is allocated to the job.
1456 no[ne] don't bind tasks to memory (default)
1458 nosort avoid sorting free cache pages (default, LaunchParameters
1460 configuration parameter can override this default)
1461 p[refer]
1463 Prefer use of first specified NUMA node, but permit
1464 use of other available NUMA nodes.
1465 q[uiet]
1467 quietly bind before task runs (default)
1468 rank bind by task rank (not recommended)
1470 sort sort free cache pages (run zonesort on Intel KNL nodes)
1472 v[erbose]
1474 verbosely report binding before task runs
1475 This option applies to job and step allocations.
1477 --mincpus=<n>
1480 Specify a minimum number of logical cpus/processors per node.
1481 This option applies to job allocations.
1482 --msg-timeout=<seconds>
1485 Modify the job launch message timeout. The default value is
1486 MessageTimeout in the Slurm configuration file slurm.conf.
1487 Changes to this are typically not recommended, but could be use‐
1488 ful to diagnose problems. This option applies to job alloca‐
1489 tions.
1490 --mpi=<mpi_type>
1493 Identify the type of MPI to be used. May result in unique initi‐
1494 ation procedures.
1495 list Lists available mpi types to choose from.
1497 openmpi
1499 For use with OpenMPI.
1500 pmi2 To enable PMI2 support. The PMI2 support in Slurm works
1502 only if the MPI implementation supports it, in other
1503 words if the MPI has the PMI2 interface implemented. The
1504 --mpi=pmi2 will load the library lib/slurm/mpi_pmi2.so
1505 which provides the server side functionality but the
1506 client side must implement PMI2_Init() and the other
1507 interface calls.
1508 pmix To enable PMIx support (http://pmix.github.io/master).
1510 The PMIx support in Slurm can be used to launch parallel
1511 applications (e.g. MPI) if it supports PMIx, PMI2 or
1512 PMI1. Slurm must be configured with pmix support by pass‐
1513 ing "--with-pmix=<PMIx installation path>" option to its
1514 "./configure" script.
1515 At the time of writing PMIx is supported in Open MPI
1517 starting from version 2.0. PMIx also supports backward
1518 compatibility with PMI1 and PMI2 and can be used if MPI
1519 was configured with PMI2/PMI1 support pointing to the
1520 PMIx library ("libpmix"). If MPI supports PMI1/PMI2 but
1521 doesn't provide the way to point to a specific implemen‐
1522 tation, a hack'ish solution leveraging LD_PRELOAD can be
1523 used to force "libpmix" usage.
1524 none No special MPI processing. This is the default and works
1527 with many other versions of MPI.
1528 This option applies to step allocations.
1530 --multi-prog
1533 Run a job with different programs and different arguments for
1534 each task. In this case, the executable program specified is
1535 actually a configuration file specifying the executable and
1536 arguments for each task. See MULTIPLE PROGRAM CONFIGURATION
1537 below for details on the configuration file contents. This
1538 option applies to step allocations.
1539 -N, --nodes=<minnodes[-maxnodes]>
1542 Request that a minimum of minnodes nodes be allocated to this
1543 job. A maximum node count may also be specified with maxnodes.
1544 If only one number is specified, this is used as both the mini‐
1545 mum and maximum node count. The partition's node limits super‐
1546 sede those of the job. If a job's node limits are outside of
1547 the range permitted for its associated partition, the job will
1548 be left in a PENDING state. This permits possible execution at
1549 a later time, when the partition limit is changed. If a job
1550 node limit exceeds the number of nodes configured in the parti‐
1551 tion, the job will be rejected. Note that the environment vari‐
1552 able SLURM_JOB_NUM_NODES (and SLURM_NNODES for backwards compat‐
1553 ibility) will be set to the count of nodes actually allocated to
1554 the job. See the ENVIRONMENT VARIABLES section for more informa‐
1555 tion. If -N is not specified, the default behavior is to allo‐
1556 cate enough nodes to satisfy the requirements of the -n and -c
1557 options. The job will be allocated as many nodes as possible
1558 within the range specified and without delaying the initiation
1559 of the job. If number of tasks is given and a number of
1560 requested nodes is also given the number of nodes used from that
1561 request will be reduced to match that of the number of tasks if
1562 the number of nodes in the request is greater than the number of
1563 tasks. The node count specification may include a numeric value
1564 followed by a suffix of "k" (multiplies numeric value by 1,024)
1565 or "m" (multiplies numeric value by 1,048,576). This option
1566 applies to job and step allocations.
1567 -n, --ntasks=<number>
1570 Specify the number of tasks to run. Request that srun allocate
1571 resources for ntasks tasks. The default is one task per node,
1572 but note that the --cpus-per-task option will change this
1573 default. This option applies to job and step allocations.
1574 --network=<type>
1577 Specify information pertaining to the switch or network. The
1578 interpretation of type is system dependent. This option is sup‐
1579 ported when running Slurm on a Cray natively. It is used to
1580 request using Network Performance Counters. Only one value per
1581 request is valid. All options are case in-sensitive. In this
1582 configuration supported values include:
1583 system
1585 Use the system-wide network performance counters. Only
1586 nodes requested will be marked in use for the job alloca‐
1587 tion. If the job does not fill up the entire system the
1588 rest of the nodes are not able to be used by other jobs
1589 using NPC, if idle their state will appear as PerfCnts.
1590 These nodes are still available for other jobs not using
1591 NPC.
1592 blade Use the blade network performance counters. Only nodes
1594 requested will be marked in use for the job allocation.
1595 If the job does not fill up the entire blade(s) allocated
1596 to the job those blade(s) are not able to be used by other
1597 jobs using NPC, if idle their state will appear as PerfC‐
1598 nts. These nodes are still available for other jobs not
1599 using NPC.
1600 In all cases the job or step allocation request must
1603 specify the
1604 --exclusive option. Otherwise the request will be denied.
1605 Also with any of these options steps are not allowed to share
1607 blades, so resources would remain idle inside an allocation if
1608 the step running on a blade does not take up all the nodes on
1609 the blade.
1610 The network option is also supported on systems with IBM's Par‐
1612 allel Environment (PE). See IBM's LoadLeveler job command key‐
1613 word documentation about the keyword "network" for more informa‐
1614 tion. Multiple values may be specified in a comma separated
1615 list. All options are case in-sensitive. Supported values
1616 include:
1617 BULK_XFER[=<resources>]
1619 Enable bulk transfer of data using Remote Direct-
1620 Memory Access (RDMA). The optional resources speci‐
1621 fication is a numeric value which can have a suffix
1622 of "k", "K", "m", "M", "g" or "G" for kilobytes,
1623 megabytes or gigabytes. NOTE: The resources speci‐
1624 fication is not supported by the underlying IBM in‐
1625 frastructure as of Parallel Environment version 2.2
1626 and no value should be specified at this time. The
1627 devices allocated to a job must all be of the same
1628 type. The default value depends upon depends upon
1629 what hardware is available and in order of prefer‐
1630 ences is IPONLY (which is not considered in User
1631 Space mode), HFI, IB, HPCE, and KMUX.
1632 CAU=<count> Number of Collective Acceleration Units (CAU)
1634 required. Applies only to IBM Power7-IH processors.
1635 Default value is zero. Independent CAU will be
1636 allocated for each programming interface (MPI, LAPI,
1637 etc.)
1638 DEVNAME=<name>
1640 Specify the device name to use for communications
1641 (e.g. "eth0" or "mlx4_0").
1642 DEVTYPE=<type>
1644 Specify the device type to use for communications.
1645 The supported values of type are: "IB" (InfiniBand),
1646 "HFI" (P7 Host Fabric Interface), "IPONLY" (IP-Only
1647 interfaces), "HPCE" (HPC Ethernet), and "KMUX" (Ker‐
1648 nel Emulation of HPCE). The devices allocated to a
1649 job must all be of the same type. The default value
1650 depends upon depends upon what hardware is available
1651 and in order of preferences is IPONLY (which is not
1652 considered in User Space mode), HFI, IB, HPCE, and
1653 KMUX.
1654 IMMED =<count>
1656 Number of immediate send slots per window required.
1657 Applies only to IBM Power7-IH processors. Default
1658 value is zero.
1659 INSTANCES =<count>
1661 Specify number of network connections for each task
1662 on each network connection. The default instance
1663 count is 1.
1664 IPV4 Use Internet Protocol (IP) version 4 communications
1666 (default).
1667 IPV6 Use Internet Protocol (IP) version 6 communications.
1669 LAPI Use the LAPI programming interface.
1671 MPI Use the MPI programming interface. MPI is the
1673 default interface.
1674 PAMI Use the PAMI programming interface.
1676 SHMEM Use the OpenSHMEM programming interface.
1678 SN_ALL Use all available switch networks (default).
1680 SN_SINGLE Use one available switch network.
1682 UPC Use the UPC programming interface.
1684 US Use User Space communications.
1686 Some examples of network specifications:
1689 Instances=2,US,MPI,SN_ALL
1691 Create two user space connections for MPI communica‐
1692 tions on every switch network for each task.
1693 US,MPI,Instances=3,Devtype=IB
1695 Create three user space connections for MPI communi‐
1696 cations on every InfiniBand network for each task.
1697 IPV4,LAPI,SN_Single
1699 Create a IP version 4 connection for LAPI communica‐
1700 tions on one switch network for each task.
1701 Instances=2,US,LAPI,MPI
1703 Create two user space connections each for LAPI and
1704 MPI communications on every switch network for each
1705 task. Note that SN_ALL is the default option so
1706 every switch network is used. Also note that
1707 Instances=2 specifies that two connections are
1708 established for each protocol (LAPI and MPI) and
1709 each task. If there are two networks and four tasks
1710 on the node then a total of 32 connections are
1711 established (2 instances x 2 protocols x 2 networks
1712 x 4 tasks).
1713 This option applies to job and step allocations.
1715 --nice[=adjustment]
1718 Run the job with an adjusted scheduling priority within Slurm.
1719 With no adjustment value the scheduling priority is decreased by
1720 100. A negative nice value increases the priority, otherwise
1721 decreases it. The adjustment range is +/- 2147483645. Only priv‐
1722 ileged users can specify a negative adjustment.
1723 --ntasks-per-core=<ntasks>
1726 Request the maximum ntasks be invoked on each core. This option
1727 applies to the job allocation, but not to step allocations.
1728 Meant to be used with the --ntasks option. Related to
1729 --ntasks-per-node except at the core level instead of the node
1730 level. Masks will automatically be generated to bind the tasks
1731 to specific core unless --cpu-bind=none is specified. NOTE:
1732 This option is not supported unless SelectType=cons_res is con‐
1733 figured (either directly or indirectly on Cray systems) along
1734 with the node's core count.
1735 --ntasks-per-node=<ntasks>
1738 Request that ntasks be invoked on each node. If used with the
1739 --ntasks option, the --ntasks option will take precedence and
1740 the --ntasks-per-node will be treated as a maximum count of
1741 tasks per node. Meant to be used with the --nodes option. This
1742 is related to --cpus-per-task=ncpus, but does not require knowl‐
1743 edge of the actual number of cpus on each node. In some cases,
1744 it is more convenient to be able to request that no more than a
1745 specific number of tasks be invoked on each node. Examples of
1746 this include submitting a hybrid MPI/OpenMP app where only one
1747 MPI "task/rank" should be assigned to each node while allowing
1748 the OpenMP portion to utilize all of the parallelism present in
1749 the node, or submitting a single setup/cleanup/monitoring job to
1750 each node of a pre-existing allocation as one step in a larger
1751 job script. This option applies to job allocations.
1752 --ntasks-per-socket=<ntasks>
1755 Request the maximum ntasks be invoked on each socket. This
1756 option applies to the job allocation, but not to step alloca‐
1757 tions. Meant to be used with the --ntasks option. Related to
1758 --ntasks-per-node except at the socket level instead of the node
1759 level. Masks will automatically be generated to bind the tasks
1760 to specific sockets unless --cpu-bind=none is specified. NOTE:
1761 This option is not supported unless SelectType=cons_res is con‐
1762 figured (either directly or indirectly on Cray systems) along
1763 with the node's socket count.
1764 -O, --overcommit
1767 Overcommit resources. This option applies to job and step allo‐
1768 cations. When applied to job allocation, only one CPU is allo‐
1769 cated to the job per node and options used to specify the number
1770 of tasks per node, socket, core, etc. are ignored. When
1771 applied to job step allocations (the srun command when executed
1772 within an existing job allocation), this option can be used to
1773 launch more than one task per CPU. Normally, srun will not
1774 allocate more than one process per CPU. By specifying --over‐
1775 commit you are explicitly allowing more than one process per
1776 CPU. However no more than MAX_TASKS_PER_NODE tasks are permitted
1777 to execute per node. NOTE: MAX_TASKS_PER_NODE is defined in the
1778 file slurm.h and is not a variable, it is set at Slurm build
1779 time.
1780 -o, --output=<filename pattern>
1783 Specify the "filename pattern" for stdout redirection. By
1784 default in interactive mode, srun collects stdout from all tasks
1785 and sends this output via TCP/IP to the attached terminal. With
1786 --output stdout may be redirected to a file, to one file per
1787 task, or to /dev/null. See section IO Redirection below for the
1788 various forms of filename pattern. If the specified file
1789 already exists, it will be overwritten.
1790 If --error is not also specified on the command line, both std‐
1792 out and stderr will directed to the file specified by --output.
1793 This option applies to job and step allocations.
1794 --open-mode=<append|truncate>
1797 Open the output and error files using append or truncate mode as
1798 specified. For heterogeneous job steps the default value is
1799 "append". Otherwise the default value is specified by the sys‐
1800 tem configuration parameter JobFileAppend. This option applies
1801 to job and step allocations.
1802 --pack-group=<expr>
1805 Identify each job in a heterogeneous job allocation for which a
1806 step is to be created. Applies only to srun commands issued
1807 inside a salloc allocation or sbatch script. <expr> is a set of
1808 integers corresponding to one or more options indexes on the
1809 salloc or sbatch command line. Examples: "--pack-group=2",
1810 "--pack-group=0,4", "--pack-group=1,3-5". The default value is
1811 --pack-group=0.
1812 -p, --partition=<partition_names>
1815 Request a specific partition for the resource allocation. If
1816 not specified, the default behavior is to allow the slurm con‐
1817 troller to select the default partition as designated by the
1818 system administrator. If the job can use more than one parti‐
1819 tion, specify their names in a comma separate list and the one
1820 offering earliest initiation will be used with no regard given
1821 to the partition name ordering (although higher priority parti‐
1822 tions will be considered first). When the job is initiated, the
1823 name of the partition used will be placed first in the job
1824 record partition string. This option applies to job allocations.
1825 --power=<flags>
1828 Comma separated list of power management plugin options. Cur‐
1829 rently available flags include: level (all nodes allocated to
1830 the job should have identical power caps, may be disabled by the
1831 Slurm configuration option PowerParameters=job_no_level). This
1832 option applies to job allocations.
1833 --priority=<value>
1836 Request a specific job priority. May be subject to configura‐
1837 tion specific constraints. value should either be a numeric
1838 value or "TOP" (for highest possible value). Only Slurm opera‐
1839 tors and administrators can set the priority of a job. This
1840 option applies to job allocations only.
1841 --profile=<all|none|[energy[,|task[,|filesystem[,|network]]]]>
1844 enables detailed data collection by the acct_gather_profile
1845 plugin. Detailed data are typically time-series that are stored
1846 in an HDF5 file for the job or an InfluxDB database depending on
1847 the configured plugin.
1848 All All data types are collected. (Cannot be combined with
1851 other values.)
1852 None No data types are collected. This is the default.
1855 (Cannot be combined with other values.)
1856 Energy Energy data is collected.
1859 Task Task (I/O, Memory, ...) data is collected.
1862 Filesystem
1865 Filesystem data is collected.
1866 Network Network (InfiniBand) data is collected.
1869 This option applies to job and step allocations.
1872 --prolog=<executable>
1875 srun will run executable just before launching the job step.
1876 The command line arguments for executable will be the command
1877 and arguments of the job step. If executable is "none", then no
1878 srun prolog will be run. This parameter overrides the SrunProlog
1879 parameter in slurm.conf. This parameter is completely indepen‐
1880 dent from the Prolog parameter in slurm.conf. This option
1881 applies to job allocations.
1882 --propagate[=rlimit[,rlimit...]]
1885 Allows users to specify which of the modifiable (soft) resource
1886 limits to propagate to the compute nodes and apply to their
1887 jobs. If no rlimit is specified, then all resource limits will
1888 be propagated. The following rlimit names are supported by
1889 Slurm (although some options may not be supported on some sys‐
1890 tems):
1891 ALL All limits listed below (default)
1893 NONE No limits listed below
1895 AS The maximum address space for a process
1897 CORE The maximum size of core file
1899 CPU The maximum amount of CPU time
1901 DATA The maximum size of a process's data segment
1903 FSIZE The maximum size of files created. Note that if the
1905 user sets FSIZE to less than the current size of the
1906 slurmd.log, job launches will fail with a 'File size
1907 limit exceeded' error.
1908 MEMLOCK The maximum size that may be locked into memory
1910 NOFILE The maximum number of open files
1912 NPROC The maximum number of processes available
1914 RSS The maximum resident set size
1916 STACK The maximum stack size
1918 This option applies to job allocations.
1920 --pty Execute task zero in pseudo terminal mode. Implicitly sets
1923 --unbuffered. Implicitly sets --error and --output to /dev/null
1924 for all tasks except task zero, which may cause those tasks to
1925 exit immediately (e.g. shells will typically exit immediately in
1926 that situation). This option applies to step allocations.
1927 -q, --qos=<qos>
1930 Request a quality of service for the job. QOS values can be
1931 defined for each user/cluster/account association in the Slurm
1932 database. Users will be limited to their association's defined
1933 set of qos's when the Slurm configuration parameter, Account‐
1934 ingStorageEnforce, includes "qos" in it's definition. This
1935 option applies to job allocations.
1936 -Q, --quiet
1939 Suppress informational messages from srun. Errors will still be
1940 displayed. This option applies to job and step allocations.
1941 --quit-on-interrupt
1944 Quit immediately on single SIGINT (Ctrl-C). Use of this option
1945 disables the status feature normally available when srun
1946 receives a single Ctrl-C and causes srun to instead immediately
1947 terminate the running job. This option applies to step alloca‐
1948 tions.
1949 -r, --relative=<n>
1952 Run a job step relative to node n of the current allocation.
1953 This option may be used to spread several job steps out among
1954 the nodes of the current job. If -r is used, the current job
1955 step will begin at node n of the allocated nodelist, where the
1956 first node is considered node 0. The -r option is not permitted
1957 with -w or -x option and will result in a fatal error when not
1958 running within a prior allocation (i.e. when SLURM_JOB_ID is not
1959 set). The default for n is 0. If the value of --nodes exceeds
1960 the number of nodes identified with the --relative option, a
1961 warning message will be printed and the --relative option will
1962 take precedence. This option applies to step allocations.
1963 --reboot
1966 Force the allocated nodes to reboot before starting the job.
1967 This is only supported with some system configurations and will
1968 otherwise be silently ignored. This option applies to job allo‐
1969 cations.
1970 --resv-ports[=count]
1973 Reserve communication ports for this job. Users can specify the
1974 number of port they want to reserve. The parameter Mpi‐
1975 Params=ports=12000-12999 must be specified in slurm.conf. If not
1976 specified and Slurm's OpenMPI plugin is used, then by default
1977 the number of reserved equal to the highest number of tasks on
1978 any node in the job step allocation. If the number of reserved
1979 ports is zero then no ports is reserved. Used for OpenMPI. This
1980 option applies to job and step allocations.
1981 --reservation=<name>
1984 Allocate resources for the job from the named reservation. This
1985 option applies to job allocations.
1986 -s, --oversubscribe
1989 The job allocation can over-subscribe resources with other run‐
1990 ning jobs. The resources to be over-subscribed can be nodes,
1991 sockets, cores, and/or hyperthreads depending upon configura‐
1992 tion. The default over-subscribe behavior depends on system
1993 configuration and the partition's OverSubscribe option takes
1994 precedence over the job's option. This option may result in the
1995 allocation being granted sooner than if the --oversubscribe
1996 option was not set and allow higher system utilization, but
1997 application performance will likely suffer due to competition
1998 for resources. Also see the --exclusive option. This option
1999 applies to step allocations.
2000 -S, --core-spec=<num>
2003 Count of specialized cores per node reserved by the job for sys‐
2004 tem operations and not used by the application. The application
2005 will not use these cores, but will be charged for their alloca‐
2006 tion. Default value is dependent upon the node's configured
2007 CoreSpecCount value. If a value of zero is designated and the
2008 Slurm configuration option AllowSpecResourcesUsage is enabled,
2009 the job will be allowed to override CoreSpecCount and use the
2010 specialized resources on nodes it is allocated. This option can
2011 not be used with the --thread-spec option. This option applies
2012 to job allocations.
2013 --signal=<sig_num>[@<sig_time>]
2016 When a job is within sig_time seconds of its end time, send it
2017 the signal sig_num. Due to the resolution of event handling by
2018 Slurm, the signal may be sent up to 60 seconds earlier than
2019 specified. sig_num may either be a signal number or name (e.g.
2020 "10" or "USR1"). sig_time must have an integer value between 0
2021 and 65535. By default, no signal is sent before the job's end
2022 time. If a sig_num is specified without any sig_time, the
2023 default time will be 60 seconds. This option applies to job
2024 allocations. To have the signal sent at preemption time see the
2025 preempt_send_user_signal SlurmctldParameter.
2026 --slurmd-debug=<level>
2029 Specify a debug level for slurmd(8). The level may be specified
2030 either an integer value between 0 [quiet, only errors are dis‐
2031 played] and 4 [verbose operation] or the SlurmdDebug tags.
2032 quiet Log nothing
2034 fatal Log only fatal errors
2036 error Log only errors
2038 info Log errors and general informational messages
2040 verbose Log errors and verbose informational messages
2042 The slurmd debug information is copied onto the stderr of
2045 the job. By default only errors are displayed. This option
2046 applies to job and step allocations.
2047 --sockets-per-node=<sockets>
2050 Restrict node selection to nodes with at least the specified
2051 number of sockets. See additional information under -B option
2052 above when task/affinity plugin is enabled. This option applies
2053 to job allocations.
2054 --spread-job
2057 Spread the job allocation over as many nodes as possible and
2058 attempt to evenly distribute tasks across the allocated nodes.
2059 This option disables the topology/tree plugin. This option
2060 applies to job allocations.
2061 --switches=<count>[@<max-time>]
2064 When a tree topology is used, this defines the maximum count of
2065 switches desired for the job allocation and optionally the maxi‐
2066 mum time to wait for that number of switches. If Slurm finds an
2067 allocation containing more switches than the count specified,
2068 the job remains pending until it either finds an allocation with
2069 desired switch count or the time limit expires. It there is no
2070 switch count limit, there is no delay in starting the job.
2071 Acceptable time formats include "minutes", "minutes:seconds",
2072 "hours:minutes:seconds", "days-hours", "days-hours:minutes" and
2073 "days-hours:minutes:seconds". The job's maximum time delay may
2074 be limited by the system administrator using the SchedulerParam‐
2075 eters configuration parameter with the max_switch_wait parameter
2076 option. On a dragonfly network the only switch count supported
2077 is 1 since communication performance will be highest when a job
2078 is allocate resources on one leaf switch or more than 2 leaf
2079 switches. The default max-time is the max_switch_wait Sched‐
2080 ulerParameters. This option applies to job allocations.
2081 -T, --threads=<nthreads>
2084 Allows limiting the number of concurrent threads used to send
2085 the job request from the srun process to the slurmd processes on
2086 the allocated nodes. Default is to use one thread per allocated
2087 node up to a maximum of 60 concurrent threads. Specifying this
2088 option limits the number of concurrent threads to nthreads (less
2089 than or equal to 60). This should only be used to set a low
2090 thread count for testing on very small memory computers. This
2091 option applies to job allocations.
2092 -t, --time=<time>
2095 Set a limit on the total run time of the job allocation. If the
2096 requested time limit exceeds the partition's time limit, the job
2097 will be left in a PENDING state (possibly indefinitely). The
2098 default time limit is the partition's default time limit. When
2099 the time limit is reached, each task in each job step is sent
2100 SIGTERM followed by SIGKILL. The interval between signals is
2101 specified by the Slurm configuration parameter KillWait. The
2102 OverTimeLimit configuration parameter may permit the job to run
2103 longer than scheduled. Time resolution is one minute and second
2104 values are rounded up to the next minute.
2105 A time limit of zero requests that no time limit be imposed.
2107 Acceptable time formats include "minutes", "minutes:seconds",
2108 "hours:minutes:seconds", "days-hours", "days-hours:minutes" and
2109 "days-hours:minutes:seconds". This option applies to job and
2110 step allocations.
2111 --task-epilog=<executable>
2114 The slurmstepd daemon will run executable just after each task
2115 terminates. This will be executed before any TaskEpilog parame‐
2116 ter in slurm.conf is executed. This is meant to be a very
2117 short-lived program. If it fails to terminate within a few sec‐
2118 onds, it will be killed along with any descendant processes.
2119 This option applies to step allocations.
2120 --task-prolog=<executable>
2123 The slurmstepd daemon will run executable just before launching
2124 each task. This will be executed after any TaskProlog parameter
2125 in slurm.conf is executed. Besides the normal environment vari‐
2126 ables, this has SLURM_TASK_PID available to identify the process
2127 ID of the task being started. Standard output from this program
2128 of the form "export NAME=value" will be used to set environment
2129 variables for the task being spawned. This option applies to
2130 step allocations.
2131 --test-only
2134 Returns an estimate of when a job would be scheduled to run
2135 given the current job queue and all the other srun arguments
2136 specifying the job. This limits srun's behavior to just return
2137 information; no job is actually submitted. The program will be
2138 executed directly by the slurmd daemon. This option applies to
2139 job allocations.
2140 --thread-spec=<num>
2143 Count of specialized threads per node reserved by the job for
2144 system operations and not used by the application. The applica‐
2145 tion will not use these threads, but will be charged for their
2146 allocation. This option can not be used with the --core-spec
2147 option. This option applies to job allocations.
2148 --threads-per-core=<threads>
2151 Restrict node selection to nodes with at least the specified
2152 number of threads per core. NOTE: "Threads" refers to the num‐
2153 ber of processing units on each core rather than the number of
2154 application tasks to be launched per core. See additional
2155 information under -B option above when task/affinity plugin is
2156 enabled. This option applies to job allocations.
2157 --time-min=<time>
2160 Set a minimum time limit on the job allocation. If specified,
2161 the job may have it's --time limit lowered to a value no lower
2162 than --time-min if doing so permits the job to begin execution
2163 earlier than otherwise possible. The job's time limit will not
2164 be changed after the job is allocated resources. This is per‐
2165 formed by a backfill scheduling algorithm to allocate resources
2166 otherwise reserved for higher priority jobs. Acceptable time
2167 formats include "minutes", "minutes:seconds", "hours:min‐
2168 utes:seconds", "days-hours", "days-hours:minutes" and
2169 "days-hours:minutes:seconds". This option applies to job alloca‐
2170 tions.
2171 --tmp=<size[units]>
2174 Specify a minimum amount of temporary disk space per node.
2175 Default units are megabytes unless the SchedulerParameters con‐
2176 figuration parameter includes the "default_gbytes" option for
2177 gigabytes. Different units can be specified using the suffix
2178 [K|M|G|T]. This option applies to job allocations.
2179 -u, --unbuffered
2182 By default the connection between slurmstepd and the user
2183 launched application is over a pipe. The stdio output written by
2184 the application is buffered by the glibc until it is flushed or
2185 the output is set as unbuffered. See setbuf(3). If this option
2186 is specified the tasks are executed with a pseudo terminal so
2187 that the application output is unbuffered. This option applies
2188 to step allocations.
2189 --usage
2191 Display brief help message and exit.
2192 --uid=<user>
2195 Attempt to submit and/or run a job as user instead of the invok‐
2196 ing user id. The invoking user's credentials will be used to
2197 check access permissions for the target partition. User root may
2198 use this option to run jobs as a normal user in a RootOnly par‐
2199 tition for example. If run as root, srun will drop its permis‐
2200 sions to the uid specified after node allocation is successful.
2201 user may be the user name or numerical user ID. This option
2202 applies to job and step allocations.
2203 --use-min-nodes
2206 If a range of node counts is given, prefer the smaller count.
2207 -V, --version
2210 Display version information and exit.
2211 -v, --verbose
2214 Increase the verbosity of srun's informational messages. Multi‐
2215 ple -v's will further increase srun's verbosity. By default
2216 only errors will be displayed. This option applies to job and
2217 step allocations.
2218 -W, --wait=<seconds>
2221 Specify how long to wait after the first task terminates before
2222 terminating all remaining tasks. A value of 0 indicates an
2223 unlimited wait (a warning will be issued after 60 seconds). The
2224 default value is set by the WaitTime parameter in the slurm con‐
2225 figuration file (see slurm.conf(5)). This option can be useful
2226 to ensure that a job is terminated in a timely fashion in the
2227 event that one or more tasks terminate prematurely. Note: The
2228 -K, --kill-on-bad-exit option takes precedence over -W, --wait
2229 to terminate the job immediately if a task exits with a non-zero
2230 exit code. This option applies to job allocations.
2231 -w, --nodelist=<host1,host2,... or filename>
2234 Request a specific list of hosts. The job will contain all of
2235 these hosts and possibly additional hosts as needed to satisfy
2236 resource requirements. The list may be specified as a
2237 comma-separated list of hosts, a range of hosts (host[1-5,7,...]
2238 for example), or a filename. The host list will be assumed to
2239 be a filename if it contains a "/" character. If you specify a
2240 minimum node or processor count larger than can be satisfied by
2241 the supplied host list, additional resources will be allocated
2242 on other nodes as needed. Rather than repeating a host name
2243 multiple times, an asterisk and a repetition count may be
2244 appended to a host name. For example "host1,host1" and "host1*2"
2245 are equivalent. If number of tasks is given and a list of
2246 requested nodes is also given the number of nodes used from that
2247 list will be reduced to match that of the number of tasks if the
2248 number of nodes in the list is greater than the number of tasks.
2249 This option applies to job and step allocations.
2250 --wckey=<wckey>
2253 Specify wckey to be used with job. If TrackWCKey=no (default)
2254 in the slurm.conf this value is ignored. This option applies to
2255 job allocations.
2256 -X, --disable-status
2259 Disable the display of task status when srun receives a single
2260 SIGINT (Ctrl-C). Instead immediately forward the SIGINT to the
2261 running job. Without this option a second Ctrl-C in one second
2262 is required to forcibly terminate the job and srun will immedi‐
2263 ately exit. May also be set via the environment variable
2264 SLURM_DISABLE_STATUS. This option applies to job allocations.
2265 -x, --exclude=<host1,host2,... or filename>
2268 Request that a specific list of hosts not be included in the
2269 resources allocated to this job. The host list will be assumed
2270 to be a filename if it contains a "/"character. This option
2271 applies to job allocations.
2272 --x11[=<all|first|last>]
2275 Sets up X11 forwarding on all, first or last node(s) of the
2276 allocation. This option is only enabled if Slurm was compiled
2277 with X11 support and PrologFlags=x11 is defined in the
2278 slurm.conf. Default is all.
2279 -Z, --no-allocate
2282 Run the specified tasks on a set of nodes without creating a
2283 Slurm "job" in the Slurm queue structure, bypassing the normal
2284 resource allocation step. The list of nodes must be specified
2285 with the -w, --nodelist option. This is a privileged option
2286 only available for the users "SlurmUser" and "root". This option
2287 applies to job allocations.
2288 srun will submit the job request to the slurm job controller, then ini‐
2291 tiate all processes on the remote nodes. If the request cannot be met
2292 immediately, srun will block until the resources are free to run the
2293 job. If the -I (--immediate) option is specified srun will terminate if
2294 resources are not immediately available.
2295 When initiating remote processes srun will propagate the current work‐
2297 ing directory, unless --chdir=<path> is specified, in which case path
2298 will become the working directory for the remote processes.
2299 The -n, -c, and -N options control how CPUs and nodes will be allo‐
2301 cated to the job. When specifying only the number of processes to run
2302 with -n, a default of one CPU per process is allocated. By specifying
2303 the number of CPUs required per task (-c), more than one CPU may be
2304 allocated per process. If the number of nodes is specified with -N,
2305 srun will attempt to allocate at least the number of nodes specified.
2306 Combinations of the above three options may be used to change how pro‐
2308 cesses are distributed across nodes and cpus. For instance, by specify‐
2309 ing both the number of processes and number of nodes on which to run,
2310 the number of processes per node is implied. However, if the number of
2311 CPUs per process is more important then number of processes (-n) and
2312 the number of CPUs per process (-c) should be specified.
2313 srun will refuse to allocate more than one process per CPU unless
2315 --overcommit (-O) is also specified.
2316 srun will attempt to meet the above specifications "at a minimum." That
2318 is, if 16 nodes are requested for 32 processes, and some nodes do not
2319 have 2 CPUs, the allocation of nodes will be increased in order to meet
2320 the demand for CPUs. In other words, a minimum of 16 nodes are being
2321 requested. However, if 16 nodes are requested for 15 processes, srun
2322 will consider this an error, as 15 processes cannot run across 16
2323 nodes.
2324 IO Redirection
2327 By default, stdout and stderr will be redirected from all tasks to the
2329 stdout and stderr of srun, and stdin will be redirected from the stan‐
2330 dard input of srun to all remote tasks. If stdin is only to be read by
2331 a subset of the spawned tasks, specifying a file to read from rather
2332 than forwarding stdin from the srun command may be preferable as it
2333 avoids moving and storing data that will never be read.
2334 For OS X, the poll() function does not support stdin, so input from a
2336 terminal is not possible.
2337 This behavior may be changed with the --output, --error, and --input
2339 (-o, -e, -i) options. Valid format specifications for these options are
2340 all stdout stderr is redirected from all tasks to srun. stdin is
2342 broadcast to all remote tasks. (This is the default behav‐
2343 ior)
2344 none stdout and stderr is not received from any task. stdin is
2346 not sent to any task (stdin is closed).
2347 taskid stdout and/or stderr are redirected from only the task with
2349 relative id equal to taskid, where 0 <= taskid <= ntasks,
2350 where ntasks is the total number of tasks in the current job
2351 step. stdin is redirected from the stdin of srun to this
2352 same task. This file will be written on the node executing
2353 the task.
2354 filename srun will redirect stdout and/or stderr to the named file
2356 from all tasks. stdin will be redirected from the named file
2357 and broadcast to all tasks in the job. filename refers to a
2358 path on the host that runs srun. Depending on the cluster's
2359 file system layout, this may result in the output appearing
2360 in different places depending on whether the job is run in
2361 batch mode.
2362 filename pattern
2364 srun allows for a filename pattern to be used to generate the
2365 named IO file described above. The following list of format
2366 specifiers may be used in the format string to generate a
2367 filename that will be unique to a given jobid, stepid, node,
2368 or task. In each case, the appropriate number of files are
2369 opened and associated with the corresponding tasks. Note that
2370 any format string containing %t, %n, and/or %N will be writ‐
2371 ten on the node executing the task rather than the node where
2372 srun executes, these format specifiers are not supported on a
2373 BGQ system.
2374 \\ Do not process any of the replacement symbols.
2376 %% The character "%".
2378 %A Job array's master job allocation number.
2380 %a Job array ID (index) number.
2382 %J jobid.stepid of the running job. (e.g. "128.0")
2384 %j jobid of the running job.
2386 %s stepid of the running job.
2388 %N short hostname. This will create a separate IO file
2390 per node.
2391 %n Node identifier relative to current job (e.g. "0" is
2393 the first node of the running job) This will create a
2394 separate IO file per node.
2395 %t task identifier (rank) relative to current job. This
2397 will create a separate IO file per task.
2398 %u User name.
2400 %x Job name.
2402 A number placed between the percent character and format
2404 specifier may be used to zero-pad the result in the IO file‐
2405 name. This number is ignored if the format specifier corre‐
2406 sponds to non-numeric data (%N for example).
2407 Some examples of how the format string may be used for a 4
2409 task job step with a Job ID of 128 and step id of 0 are
2410 included below:
2411 job%J.out job128.0.out
2413 job%4j.out job0128.out
2415 job%j-%2t.out job128-00.out, job128-01.out, ...
2417
2419 Some srun options may be set via environment variables. These environ‐
2420 ment variables, along with their corresponding options, are listed
2421 below. Note: Command line options will always override these settings.
2422 PMI_FANOUT This is used exclusively with PMI (MPICH2 and
2424 MVAPICH2) and controls the fanout of data commu‐
2425 nications. The srun command sends messages to
2426 application programs (via the PMI library) and
2427 those applications may be called upon to forward
2428 that data to up to this number of additional
2429 tasks. Higher values offload work from the srun
2430 command to the applications and likely increase
2431 the vulnerability to failures. The default value
2432 is 32.
2433 PMI_FANOUT_OFF_HOST This is used exclusively with PMI (MPICH2 and
2435 MVAPICH2) and controls the fanout of data commu‐
2436 nications. The srun command sends messages to
2437 application programs (via the PMI library) and
2438 those applications may be called upon to forward
2439 that data to additional tasks. By default, srun
2440 sends one message per host and one task on that
2441 host forwards the data to other tasks on that
2442 host up to PMI_FANOUT. If PMI_FANOUT_OFF_HOST is
2443 defined, the user task may be required to forward
2444 the data to tasks on other hosts. Setting
2445 PMI_FANOUT_OFF_HOST may increase performance.
2446 Since more work is performed by the PMI library
2447 loaded by the user application, failures also can
2448 be more common and more difficult to diagnose.
2449 PMI_TIME This is used exclusively with PMI (MPICH2 and
2451 MVAPICH2) and controls how much the communica‐
2452 tions from the tasks to the srun are spread out
2453 in time in order to avoid overwhelming the srun
2454 command with work. The default value is 500
2455 (microseconds) per task. On relatively slow pro‐
2456 cessors or systems with very large processor
2457 counts (and large PMI data sets), higher values
2458 may be required.
2459 SLURM_CONF The location of the Slurm configuration file.
2461 SLURM_ACCOUNT Same as -A, --account
2463 SLURM_ACCTG_FREQ Same as --acctg-freq
2465 SLURM_BCAST Same as --bcast
2467 SLURM_BURST_BUFFER Same as --bb
2469 SLURM_CHECKPOINT Same as --checkpoint
2471 SLURM_COMPRESS Same as --compress
2473 SLURM_CONSTRAINT Same as -C, --constraint
2475 SLURM_CORE_SPEC Same as --core-spec
2477 SLURM_CPU_BIND Same as --cpu-bind
2479 SLURM_CPU_FREQ_REQ Same as --cpu-freq.
2481 SLURM_CPUS_PER_GPU Same as --cpus-per-gpu
2483 SLURM_CPUS_PER_TASK Same as -c, --cpus-per-task
2485 SLURM_DEBUG Same as -v, --verbose
2487 SLURM_DELAY_BOOT Same as --delay-boot
2489 SLURMD_DEBUG Same as -d, --slurmd-debug
2491 SLURM_DEPENDENCY Same as -P, --dependency=<jobid>
2493 SLURM_DISABLE_STATUS Same as -X, --disable-status
2495 SLURM_DIST_PLANESIZE Same as -m plane
2497 SLURM_DISTRIBUTION Same as -m, --distribution
2499 SLURM_EPILOG Same as --epilog
2501 SLURM_EXCLUSIVE Same as --exclusive
2503 SLURM_EXIT_ERROR Specifies the exit code generated when a Slurm
2505 error occurs (e.g. invalid options). This can be
2506 used by a script to distinguish application exit
2507 codes from various Slurm error conditions. Also
2508 see SLURM_EXIT_IMMEDIATE.
2509 SLURM_EXIT_IMMEDIATE Specifies the exit code generated when the
2511 --immediate option is used and resources are not
2512 currently available. This can be used by a
2513 script to distinguish application exit codes from
2514 various Slurm error conditions. Also see
2515 SLURM_EXIT_ERROR.
2516 SLURM_EXPORT_ENV Same as --export
2518 SLURM_GPUS Same as -G, --gpus
2520 SLURM_GPU_BIND Same as --gpu-bind
2522 SLURM_GPU_FREQ Same as --gpu-freq
2524 SLURM_GPUS_PER_NODE Same as --gpus-per-node
2526 SLURM_GPUS_PER_TASK Same as --gpus-per-task
2528 SLURM_GRES_FLAGS Same as --gres-flags
2530 SLURM_HINT Same as --hint
2532 SLURM_GRES Same as --gres. Also see SLURM_STEP_GRES
2534 SLURM_IMMEDIATE Same as -I, --immediate
2536 SLURM_JOB_ID Same as --jobid
2538 SLURM_JOB_NAME Same as -J, --job-name except within an existing
2540 allocation, in which case it is ignored to avoid
2541 using the batch job's name as the name of each
2542 job step.
2543 SLURM_JOB_NUM_NODES (and SLURM_NNODES for backwards compatibility)
2545 Same as -N, --nodes Total number of nodes in the
2546 job’s resource allocation.
2547 SLURM_KILL_BAD_EXIT Same as -K, --kill-on-bad-exit
2549 SLURM_LABELIO Same as -l, --label
2551 SLURM_MEM_BIND Same as --mem-bind
2553 SLURM_MEM_PER_CPU Same as --mem-per-cpu
2555 SLURM_MEM_PER_GPU Same as --mem-per-gpu
2557 SLURM_MEM_PER_NODE Same as --mem
2559 SLURM_MPI_TYPE Same as --mpi
2561 SLURM_NETWORK Same as --network
2563 SLURM_NO_KILL Same as -k, --no-kill
2565 SLURM_NTASKS (and SLURM_NPROCS for backwards compatibility)
2567 Same as -n, --ntasks
2568 SLURM_NTASKS_PER_CORE Same as --ntasks-per-core
2570 SLURM_NTASKS_PER_NODE Same as --ntasks-per-node
2572 SLURM_NTASKS_PER_SOCKET
2574 Same as --ntasks-per-socket
2575 SLURM_OPEN_MODE Same as --open-mode
2577 SLURM_OVERCOMMIT Same as -O, --overcommit
2579 SLURM_PARTITION Same as -p, --partition
2581 SLURM_PMI_KVS_NO_DUP_KEYS
2583 If set, then PMI key-pairs will contain no dupli‐
2584 cate keys. MPI can use this variable to inform
2585 the PMI library that it will not use duplicate
2586 keys so PMI can skip the check for duplicate
2587 keys. This is the case for MPICH2 and reduces
2588 overhead in testing for duplicates for improved
2589 performance
2590 SLURM_POWER Same as --power
2592 SLURM_PROFILE Same as --profile
2594 SLURM_PROLOG Same as --prolog
2596 SLURM_QOS Same as --qos
2598 SLURM_REMOTE_CWD Same as -D, --chdir=
2600 SLURM_REQ_SWITCH When a tree topology is used, this defines the
2602 maximum count of switches desired for the job
2603 allocation and optionally the maximum time to
2604 wait for that number of switches. See --switches
2605 SLURM_RESERVATION Same as --reservation
2607 SLURM_RESV_PORTS Same as --resv-ports
2609 SLURM_SIGNAL Same as --signal
2611 SLURM_STDERRMODE Same as -e, --error
2613 SLURM_STDINMODE Same as -i, --input
2615 SLURM_SPREAD_JOB Same as --spread-job
2617 SLURM_SRUN_REDUCE_TASK_EXIT_MSG
2619 if set and non-zero, successive task exit mes‐
2620 sages with the same exit code will be printed
2621 only once.
2622 SLURM_STEP_GRES Same as --gres (only applies to job steps, not to
2624 job allocations). Also see SLURM_GRES
2625 SLURM_STEP_KILLED_MSG_NODE_ID=ID
2627 If set, only the specified node will log when the
2628 job or step are killed by a signal.
2629 SLURM_STDOUTMODE Same as -o, --output
2631 SLURM_TASK_EPILOG Same as --task-epilog
2633 SLURM_TASK_PROLOG Same as --task-prolog
2635 SLURM_TEST_EXEC If defined, srun will verify existence of the
2637 executable program along with user execute per‐
2638 mission on the node where srun was called before
2639 attempting to launch it on nodes in the step.
2640 SLURM_THREAD_SPEC Same as --thread-spec
2642 SLURM_THREADS Same as -T, --threads
2644 SLURM_TIMELIMIT Same as -t, --time
2646 SLURM_UNBUFFEREDIO Same as -u, --unbuffered
2648 SLURM_USE_MIN_NODES Same as --use-min-nodes
2650 SLURM_WAIT Same as -W, --wait
2652 SLURM_WAIT4SWITCH Max time waiting for requested switches. See
2654 --switches
2655 SLURM_WCKEY Same as -W, --wckey
2657 SLURM_WORKING_DIR -D, --chdir
2659 SRUN_EXPORT_ENV Same as --export, and will override any setting
2661 for SRUN_EXPORT_ENV.
2662
2666 srun will set some environment variables in the environment of the exe‐
2667 cuting tasks on the remote compute nodes. These environment variables
2668 are:
2669 SLURM_*_PACK_GROUP_# For a heterogeneous job allocation, the environ‐
2672 ment variables are set separately for each compo‐
2673 nent.
2674 SLURM_CLUSTER_NAME Name of the cluster on which the job is execut‐
2676 ing.
2677 SLURM_CPU_BIND_VERBOSE
2679 --cpu-bind verbosity (quiet,verbose).
2680 SLURM_CPU_BIND_TYPE --cpu-bind type (none,rank,map_cpu:,mask_cpu:).
2682 SLURM_CPU_BIND_LIST --cpu-bind map or mask list (list of Slurm CPU
2684 IDs or masks for this node, CPU_ID = Board_ID x
2685 threads_per_board + Socket_ID x
2686 threads_per_socket + Core_ID x threads_per_core +
2687 Thread_ID).
2688 SLURM_CPU_FREQ_REQ Contains the value requested for cpu frequency on
2691 the srun command as a numerical frequency in
2692 kilohertz, or a coded value for a request of low,
2693 medium,highm1 or high for the frequency. See the
2694 description of the --cpu-freq option or the
2695 SLURM_CPU_FREQ_REQ input environment variable.
2696 SLURM_CPUS_ON_NODE Count of processors available to the job on this
2698 node. Note the select/linear plugin allocates
2699 entire nodes to jobs, so the value indicates the
2700 total count of CPUs on the node. For the
2701 select/cons_res plugin, this number indicates the
2702 number of cores on this node allocated to the
2703 job.
2704 SLURM_CPUS_PER_GPU Number of CPUs requested per allocated GPU. Only
2706 set if the --cpus-per-gpu option is specified.
2707 SLURM_CPUS_PER_TASK Number of cpus requested per task. Only set if
2709 the --cpus-per-task option is specified.
2710 SLURM_DISTRIBUTION Distribution type for the allocated jobs. Set the
2712 distribution with -m, --distribution.
2713 SLURM_GPUS Number of GPUs requested. Only set if the -G,
2715 --gpus option is specified.
2716 SLURM_GPU_BIND Requested binding of tasks to GPU. Only set if
2718 the --gpu-bind option is specified.
2719 SLURM_GPU_FREQ Requested GPU frequency. Only set if the
2721 --gpu-freq option is specified.
2722 SLURM_GPUS_PER_NODE Requested GPU count per allocated node. Only set
2724 if the --gpus-per-node option is specified.
2725 SLURM_GPUS_PER_SOCKET Requested GPU count per allocated socket. Only
2727 set if the --gpus-per-socket option is specified.
2728 SLURM_GPUS_PER_TASK Requested GPU count per allocated task. Only set
2730 if the --gpus-per-task option is specified.
2731 SLURM_GTIDS Global task IDs running on this node. Zero ori‐
2733 gin and comma separated.
2734 SLURM_JOB_ACCOUNT Account name associated of the job allocation.
2736 SLURM_JOB_CPUS_PER_NODE
2738 Number of CPUS per node.
2739 SLURM_JOB_DEPENDENCY Set to value of the --dependency option.
2741 SLURM_JOB_ID (and SLURM_JOBID for backwards compatibility)
2743 Job id of the executing job.
2744 SLURM_JOB_NAME Set to the value of the --job-name option or the
2747 command name when srun is used to create a new
2748 job allocation. Not set when srun is used only to
2749 create a job step (i.e. within an existing job
2750 allocation).
2751 SLURM_JOB_PARTITION Name of the partition in which the job is run‐
2754 ning.
2755 SLURM_JOB_QOS Quality Of Service (QOS) of the job allocation.
2758 SLURM_JOB_RESERVATION Advanced reservation containing the job alloca‐
2760 tion, if any.
2761 SLURM_LAUNCH_NODE_IPADDR
2764 IP address of the node from which the task launch
2765 was initiated (where the srun command ran from).
2766 SLURM_LOCALID Node local task ID for the process within a job.
2768 SLURM_MEM_BIND_LIST --mem-bind map or mask list (<list of IDs or
2771 masks for this node>).
2772 SLURM_MEM_BIND_PREFER --mem-bind prefer (prefer).
2774 SLURM_MEM_BIND_SORT Sort free cache pages (run zonesort on Intel KNL
2776 nodes).
2777 SLURM_MEM_BIND_TYPE --mem-bind type (none,rank,map_mem:,mask_mem:).
2779 SLURM_MEM_BIND_VERBOSE
2781 --mem-bind verbosity (quiet,verbose).
2782 SLURM_MEM_PER_GPU Requested memory per allocated GPU. Only set if
2784 the --mem-per-gpu option is specified.
2785 SLURM_JOB_NODES Total number of nodes in the job's resource allo‐
2787 cation.
2788 SLURM_NODE_ALIASES Sets of node name, communication address and
2790 hostname for nodes allocated to the job from the
2791 cloud. Each element in the set if colon separated
2792 and each set is comma separated. For example:
2793 SLURM_NODE_ALIASES=
2794 ec0:1.2.3.4:foo,ec1:1.2.3.5:bar
2795 SLURM_NODEID The relative node ID of the current node.
2797 SLURM_JOB_NODELIST List of nodes allocated to the job.
2799 SLURM_NTASKS (and SLURM_NPROCS for backwards compatibility)
2801 Total number of processes in the current job or
2802 job step.
2803 SLURM_PACK_SIZE Set to count of components in heterogeneous job.
2805 SLURM_PRIO_PROCESS The scheduling priority (nice value) at the time
2807 of job submission. This value is propagated to
2808 the spawned processes.
2809 SLURM_PROCID The MPI rank (or relative process ID) of the cur‐
2811 rent process.
2812 SLURM_SRUN_COMM_HOST IP address of srun communication host.
2814 SLURM_SRUN_COMM_PORT srun communication port.
2816 SLURM_STEP_LAUNCHER_PORT
2818 Step launcher port.
2819 SLURM_STEP_NODELIST List of nodes allocated to the step.
2821 SLURM_STEP_NUM_NODES Number of nodes allocated to the step.
2823 SLURM_STEP_NUM_TASKS Number of processes in the step.
2825 SLURM_STEP_TASKS_PER_NODE
2827 Number of processes per node within the step.
2828 SLURM_STEP_ID (and SLURM_STEPID for backwards compatibility)
2830 The step ID of the current job.
2831 SLURM_SUBMIT_DIR The directory from which srun was invoked or, if
2833 applicable, the directory specified by the -D,
2834 --chdir option.
2835 SLURM_SUBMIT_HOST The hostname of the computer from which salloc
2837 was invoked.
2838 SLURM_TASK_PID The process ID of the task being started.
2840 SLURM_TASKS_PER_NODE Number of tasks to be initiated on each node.
2842 Values are comma separated and in the same order
2843 as SLURM_JOB_NODELIST. If two or more consecu‐
2844 tive nodes are to have the same task count, that
2845 count is followed by "(x#)" where "#" is the rep‐
2846 etition count. For example,
2847 "SLURM_TASKS_PER_NODE=2(x3),1" indicates that the
2848 first three nodes will each execute three tasks
2849 and the fourth node will execute one task.
2850 SLURM_TOPOLOGY_ADDR This is set only if the system has the topol‐
2853 ogy/tree plugin configured. The value will be
2854 set to the names network switches which may be
2855 involved in the job's communications from the
2856 system's top level switch down to the leaf switch
2857 and ending with node name. A period is used to
2858 separate each hardware component name.
2859 SLURM_TOPOLOGY_ADDR_PATTERN
2861 This is set only if the system has the topol‐
2862 ogy/tree plugin configured. The value will be
2863 set component types listed in SLURM_TOPOL‐
2864 OGY_ADDR. Each component will be identified as
2865 either "switch" or "node". A period is used to
2866 separate each hardware component type.
2867 SLURM_UMASK The umask in effect when the job was submitted.
2869 SLURMD_NODENAME Name of the node running the task. In the case of
2871 a parallel job executing on multiple compute
2872 nodes, the various tasks will have this environ‐
2873 ment variable set to different values on each
2874 compute node.
2875 SRUN_DEBUG Set to the logging level of the srun command.
2877 Default value is 3 (info level). The value is
2878 incremented or decremented based upon the --ver‐
2879 bose and --quiet options.
2880
2883 Signals sent to the srun command are automatically forwarded to the
2884 tasks it is controlling with a few exceptions. The escape sequence
2885 <control-c> will report the state of all tasks associated with the srun
2886 command. If <control-c> is entered twice within one second, then the
2887 associated SIGINT signal will be sent to all tasks and a termination
2888 sequence will be entered sending SIGCONT, SIGTERM, and SIGKILL to all
2889 spawned tasks. If a third <control-c> is received, the srun program
2890 will be terminated without waiting for remote tasks to exit or their
2891 I/O to complete.
2892 The escape sequence <control-z> is presently ignored. Our intent is for
2894 this put the srun command into a mode where various special actions may
2895 be invoked.
2896
2899 MPI use depends upon the type of MPI being used. There are three fun‐
2900 damentally different modes of operation used by these various MPI
2901 implementation.
2902 1. Slurm directly launches the tasks and performs initialization of
2904 communications through the PMI2 or PMIx APIs. For example: "srun -n16
2905 a.out".
2906 2. Slurm creates a resource allocation for the job and then mpirun
2908 launches tasks using Slurm's infrastructure (OpenMPI).
2909 3. Slurm creates a resource allocation for the job and then mpirun
2911 launches tasks using some mechanism other than Slurm, such as SSH or
2912 RSH. These tasks are initiated outside of Slurm's monitoring or con‐
2913 trol. Slurm's epilog should be configured to purge these tasks when the
2914 job's allocation is relinquished, or the use of pam_slurm_adopt is
2915 highly recommended.
2916 See https://slurm.schedmd.com/mpi_guide.html for more information on
2918 use of these various MPI implementation with Slurm.
2919
2922 Comments in the configuration file must have a "#" in column one. The
2923 configuration file contains the following fields separated by white
2924 space:
2925 Task rank
2927 One or more task ranks to use this configuration. Multiple val‐
2928 ues may be comma separated. Ranges may be indicated with two
2929 numbers separated with a '-' with the smaller number first (e.g.
2930 "0-4" and not "4-0"). To indicate all tasks not otherwise spec‐
2931 ified, specify a rank of '*' as the last line of the file. If
2932 an attempt is made to initiate a task for which no executable
2933 program is defined, the following error message will be produced
2934 "No executable program specified for this task".
2935 Executable
2937 The name of the program to execute. May be fully qualified
2938 pathname if desired.
2939 Arguments
2941 Program arguments. The expression "%t" will be replaced with
2942 the task's number. The expression "%o" will be replaced with
2943 the task's offset within this range (e.g. a configured task rank
2944 value of "1-5" would have offset values of "0-4"). Single
2945 quotes may be used to avoid having the enclosed values inter‐
2946 preted. This field is optional. Any arguments for the program
2947 entered on the command line will be added to the arguments spec‐
2948 ified in the configuration file.
2949 For example:
2951 ###################################################################
2952 # srun multiple program configuration file
2953 #
2954 # srun -n8 -l --multi-prog silly.conf
2955 ###################################################################
2956 4-6 hostname
2957 1,7 echo task:%t
2958 0,2-3 echo offset:%o
2959 > srun -n8 -l --multi-prog silly.conf
2961 0: offset:0
2962 1: task:1
2963 2: offset:1
2964 3: offset:2
2965 4: linux15.llnl.gov
2966 5: linux16.llnl.gov
2967 6: linux17.llnl.gov
2968 7: task:7
2969
2974 This simple example demonstrates the execution of the command hostname
2975 in eight tasks. At least eight processors will be allocated to the job
2976 (the same as the task count) on however many nodes are required to sat‐
2977 isfy the request. The output of each task will be proceeded with its
2978 task number. (The machine "dev" in the example below has a total of
2979 two CPUs per node)
2980 > srun -n8 -l hostname
2983 0: dev0
2984 1: dev0
2985 2: dev1
2986 3: dev1
2987 4: dev2
2988 5: dev2
2989 6: dev3
2990 7: dev3
2991 The srun -r option is used within a job script to run two job steps on
2994 disjoint nodes in the following example. The script is run using allo‐
2995 cate mode instead of as a batch job in this case.
2996 > cat test.sh
2999 #!/bin/sh
3000 echo $SLURM_JOB_NODELIST
3001 srun -lN2 -r2 hostname
3002 srun -lN2 hostname
3003 > salloc -N4 test.sh
3005 dev[7-10]
3006 0: dev9
3007 1: dev10
3008 0: dev7
3009 1: dev8
3010 The following script runs two job steps in parallel within an allocated
3013 set of nodes.
3014 > cat test.sh
3017 #!/bin/bash
3018 srun -lN2 -n4 -r 2 sleep 60 &
3019 srun -lN2 -r 0 sleep 60 &
3020 sleep 1
3021 squeue
3022 squeue -s
3023 wait
3024 > salloc -N4 test.sh
3026 JOBID PARTITION NAME USER ST TIME NODES NODELIST
3027 65641 batch test.sh grondo R 0:01 4 dev[7-10]
3028 STEPID PARTITION USER TIME NODELIST
3030 65641.0 batch grondo 0:01 dev[7-8]
3031 65641.1 batch grondo 0:01 dev[9-10]
3032 This example demonstrates how one executes a simple MPI job. We use
3035 srun to build a list of machines (nodes) to be used by mpirun in its
3036 required format. A sample command line and the script to be executed
3037 follow.
3038 > cat test.sh
3041 #!/bin/sh
3042 MACHINEFILE="nodes.$SLURM_JOB_ID"
3043 # Generate Machinefile for mpi such that hosts are in the same
3045 # order as if run via srun
3046 #
3047 srun -l /bin/hostname | sort -n | awk '{print $2}' > $MACHINEFILE
3048 # Run using generated Machine file:
3050 mpirun -np $SLURM_NTASKS -machinefile $MACHINEFILE mpi-app
3051 rm $MACHINEFILE
3053 > salloc -N2 -n4 test.sh
3055 This simple example demonstrates the execution of different jobs on
3058 different nodes in the same srun. You can do this for any number of
3059 nodes or any number of jobs. The executables are placed on the nodes
3060 sited by the SLURM_NODEID env var. Starting at 0 and going to the num‐
3061 ber specified on the srun commandline.
3062 > cat test.sh
3065 case $SLURM_NODEID in
3066 0) echo "I am running on "
3067 hostname ;;
3068 1) hostname
3069 echo "is where I am running" ;;
3070 esac
3071 > srun -N2 test.sh
3073 dev0
3074 is where I am running
3075 I am running on
3076 dev1
3077 This example demonstrates use of multi-core options to control layout
3080 of tasks. We request that four sockets per node and two cores per
3081 socket be dedicated to the job.
3082 > srun -N2 -B 4-4:2-2 a.out
3085 This example shows a script in which Slurm is used to provide resource
3087 management for a job by executing the various job steps as processors
3088 become available for their dedicated use.
3089 > cat my.script
3092 #!/bin/bash
3093 srun --exclusive -n4 prog1 &
3094 srun --exclusive -n3 prog2 &
3095 srun --exclusive -n1 prog3 &
3096 srun --exclusive -n1 prog4 &
3097 wait
3098 This example shows how to launch an application called "master" with
3101 one task, 8 CPUs and and 16 GB of memory (2 GB per CPU) plus another
3102 application called "slave" with 16 tasks, 1 CPU per task (the default)
3103 and 1 GB of memory per task.
3104 > srun -n1 -c16 --mem-per-cpu=1gb master : -n16 --mem-per-cpu=1gb slave
3107
3110 Copyright (C) 2006-2007 The Regents of the University of California.
3111 Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
3112 Copyright (C) 2008-2010 Lawrence Livermore National Security.
3113 Copyright (C) 2010-2015 SchedMD LLC.
3114 This file is part of Slurm, a resource management program. For
3116 details, see <https://slurm.schedmd.com/>.
3117 Slurm is free software; you can redistribute it and/or modify it under
3119 the terms of the GNU General Public License as published by the Free
3120 Software Foundation; either version 2 of the License, or (at your
3121 option) any later version.
3122 Slurm is distributed in the hope that it will be useful, but WITHOUT
3124 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
3125 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
3126 for more details.
3127
3130 salloc(1), sattach(1), sbatch(1), sbcast(1), scancel(1), scontrol(1),
3131 squeue(1), slurm.conf(5), sched_setaffinity (2), numa (3) getrlimit (2)
3132October 2019 Slurm Commands srun(1)