1scancel(1)                      Slurm Commands                      scancel(1)
2
3
4

NAME

6       scancel  -  Used to signal jobs or job steps that are under the control
7       of Slurm.
8
9

SYNOPSIS

11       scancel   [OPTIONS...]    [job_id[_array_id][.step_id]]    [job_id[_ar‐
12       ray_id][.step_id]...]
13
14

DESCRIPTION

16       scancel  is used to signal or cancel jobs, job arrays or job steps.  An
17       arbitrary number of jobs or job steps may be signaled using job  speci‐
18       fication  filters  or a space separated list of specific job and/or job
19       step IDs.  If the job ID of a job array is specified with an  array  ID
20       value  then  only that job array element will be cancelled.  If the job
21       ID of a job array is specified without an array ID value then  all  job
22       array  elements  will  be cancelled.  While a heterogeneous job is in a
23       PENDING state, only the entire job can be cancelled rather than its in‐
24       dividual  components.  A request to cancel an individual component of a
25       heterogeneous job while in a PENDING state will return an error.  After
26       the  job has begun execution, an individual component can be cancelled.
27       A job or job step can only be signaled by the owner of that job or user
28       root.  If an attempt is made by an unauthorized user to signal a job or
29       job step, an error message will be printed and the job will not be sig‐
30       naled.
31
32

OPTIONS

34       -A, --account=account
35              Restrict  the  scancel  operation  to jobs under this charge ac‐
36              count.
37
38
39       -b, --batch
40              By default, signals other than SIGKILL are not sent to the batch
41              step  (the  shell script). With this option scancel signals only
42              the batch step, but not any other steps.  This  is  useful  when
43              the  shell  script has to trap the signal and take some applica‐
44              tion defined action.  Note that most shells cannot  handle  sig‐
45              nals  while  a  command  is  running (child process of the batch
46              step), the shell use to wait until the command ends to then han‐
47              dle the signal. Children of the batch step are not signaled with
48              this option, use -f, --full instead.  NOTE:  If  used  with  -f,
49              --full,  this option ignored.  NOTE: This option is not applica‐
50              ble if step_id is specified.  NOTE: The shell  itself  may  exit
51              upon  receipt of many signals.  You may avoid this by explicitly
52              trap signals within the shell script  (e.g.  "trap  <arg>  <sig‐
53              nals>"). See the shell documentation for details.
54
55
56       -M, --clusters=<string>
57              Cluster  to  issue  commands  to. Implies --ctld.  Note that the
58              SlurmDBD must be up for this option to work properly.
59
60
61       --ctld Send the job signal request to the slurmctld daemon rather  than
62              directly to the slurmd daemons. This increases overhead, but of‐
63              fers better fault tolerance.  This is the  default  behavior  on
64              architectures  using front end nodes (e.g.  Cray ALPS computers)
65              or when the --clusters option is used.
66
67
68       -f, --full
69              By default, signals other than SIGKILL are not sent to the batch
70              step  (the  shell script). With this option scancel signals also
71              the batch script and its children processes.  Most shells cannot
72              handle  signals while a command is running (child process of the
73              batch step), the shell use to wait until  the  command  ends  to
74              then  handle  the  signal.  Unlike  -b, --batch, children of the
75              batch step are also signaled with this option.  NOTE: srun steps
76              are  also children of the batch step, so steps are also signaled
77              with this option.
78
79
80       --help Print a help message describing all scancel options.
81
82
83       -H, --hurry
84              Do not stage out any burst buffer data.
85
86
87       -i, --interactive
88              Interactive mode. Confirm each job_id.step_id before  performing
89              the cancel operation.
90
91
92       -n, --jobname=job_name, --name=job_name
93              Restrict the scancel operation to jobs with this job name.
94
95
96       -w, --nodelist=host1,host2,...
97              Cancel  any  jobs using any of the given hosts.  The list may be
98              specified as a comma-separated list of hosts, a range  of  hosts
99              (host[1-5,7,...] for example), or a filename. The host list will
100              be assumed to be a filename only if it contains a "/" character.
101
102
103       -p, --partition=partition_name
104              Restrict the scancel operation to jobs in this partition.
105
106
107       -q, --qos=qos
108              Restrict the scancel operation to jobs with this quality of ser‐
109              vice.
110
111
112       -Q, --quiet
113              Do  not  report  an  error  if the specified job is already com‐
114              pleted.  This option is incompatible with the --verbose option.
115
116
117
118       -R, --reservation=reservation_name
119              Restrict the scancel operation to  jobs  with  this  reservation
120              name.
121
122
123       --sibling=cluster_name
124              Remove an active sibling job from a federated job.
125
126
127       -s, --signal=signal_name
128              The name or number of the signal to send.  If this option is not
129              used the specified job or step will be terminated. Note. If this
130              option  is  used the signal is sent directly to the slurmd where
131              the job is running bypassing the slurmctld thus  the  job  state
132              will  not  change even if the signal is delivered to it. Use the
133              scontrol command if you want the job state change  be  known  to
134              slurmctld.
135
136
137       -t, --state=job_state_name
138              Restrict   the   scancel   operation  to  jobs  in  this  state.
139              job_state_name may have a value of either  "PENDING",  "RUNNING"
140              or "SUSPENDED".
141
142
143       --usage
144              Print a brief help message listing the scancel options.
145
146
147       -u, --user=user_name
148              Restrict the scancel operation to jobs owned by this user.
149
150
151       -v, --verbose
152              Print  additional logging. Multiple v's increase logging detail.
153              This option is incompatible with the --quiet option.
154
155
156       -V, --version
157              Print the version number of the scancel command.
158
159
160       --wckey=wckey
161              Restrict the scancel operation to jobs using this workload char‐
162              acterization key.
163
164
165       ARGUMENTS
166
167
168       job_id The Slurm job ID to be signaled.
169
170
171       step_id
172              The  step  ID of the job step to be signaled.  If not specified,
173              the operation is performed at the level of a job.
174
175              If neither --batch nor --signal are used, the entire job will be
176              terminated.
177
178              When  --batch  is  used,  the batch shell processes will be sig‐
179              naled.  The child processes of the shell will not be signaled by
180              Slurm, but the shell may forward the signal.
181
182              When  --batch  is  not  used  but --signal is used, then all job
183              steps will be signaled, but the batch script itself will not  be
184              signaled.
185
186

PERFORMANCE

188       Executing scancel sends a remote procedure call to slurmctld. If enough
189       calls from scancel or other Slurm client commands that send remote pro‐
190       cedure  calls to the slurmctld daemon come in at once, it can result in
191       a degradation of performance of the slurmctld daemon, possibly  result‐
192       ing in a denial of service.
193
194       Do not run scancel or other Slurm client commands that send remote pro‐
195       cedure calls to slurmctld from loops in shell  scripts  or  other  pro‐
196       grams.  Ensure that programs limit calls to scancel to the minimum nec‐
197       essary for the information you are trying to gather.
198
199

ENVIRONMENT VARIABLES

201       Some scancel options may be set via environment variables. These  envi‐
202       ronment  variables,  along with their corresponding options, are listed
203       below.  (Note: Command line options will  always  override  these  set‐
204       tings.)
205
206       SCANCEL_ACCOUNT     -A, --account=account
207
208       SCANCEL_BATCH       -b, --batch
209
210       SCANCEL_CTLD        --ctld
211
212       SCANCEL_FULL        -f, --full
213
214       SCANCEL_HURRY       -H, --hurry
215
216       SCANCEL_INTERACTIVE -i, --interactive
217
218       SCANCEL_NAME        -n, --name=job_name
219
220       SCANCEL_PARTITION   -p, --partition=partition_name
221
222       SCANCEL_QOS         -q, --qos=qos
223
224       SCANCEL_STATE       -t, --state=job_state_name
225
226       SCANCEL_USER        -u, --user=user_name
227
228       SCANCEL_VERBOSE     -v, --verbose
229
230       SCANCEL_WCKEY       --wckey=wckey
231
232       SLURM_CONF          The location of the Slurm configuration file.
233
234       SLURM_CLUSTERS      -M, --clusters
235
236

NOTES

238       If multiple filters are supplied (e.g. --partition and --name) only the
239       jobs satisfying all of the filtering options will be signaled.
240
241       Cancelling a job step will not result in the job being terminated.  The
242       job must be cancelled to release a resource allocation.
243
244       To  cancel  a  job,  invoke scancel without --signal option.  This will
245       send first a SIGCONT to all steps to eventually wake them  up  followed
246       by a SIGTERM, then wait the KillWait duration defined in the slurm.conf
247       file and finally if they have not  terminated  send  a  SIGKILL.   This
248       gives time for the running job/step(s) to clean up.
249
250       If  a signal value of "KILL" is sent to an entire job, this will cancel
251       the active job steps but not cancel the job itself.
252
253       On Cray systems, all signals except SIGCHLD, SIGCONT, SIGSTOP, SIGTSTP,
254       SIGTTIN,  SIGTTOU, SIGURG, or SIGWINCH cause the ALPS reservation to be
255       released.  The job however will not be terminated except in the case of
256       SIGKILL and may then be used for post processing.
257
258

AUTHORIZATION

260       When  using SlurmDBD, users who have an AdminLevel defined (Operator or
261       Admin) and users who are account coordinators are given  the  authority
262       to invoke scancel on other users jobs.
263
264

EXAMPLES

266       Send SIGTERM to steps 1 and 3 of job 1234:
267
268              $ scancel --signal=TERM 1234.1 1234.3
269
270
271       Cancel job 1234 along with all of its steps:
272
273              $ scancel 1234
274
275
276       Send  SIGKILL  to  all steps of job 1235, but do not cancel the job it‐
277       self:
278
279              $ scancel --signal=KILL 1235
280
281
282       Send SIGUSR1 to the batch shell processes of job 1236:
283
284              $ scancel --signal=USR1 --batch 1236
285
286
287       Cancel all pending jobs belonging to user "bob" in partition "debug":
288
289              $ scancel --state=PENDING --user=bob --partition=debug
290
291
292       Cancel only array ID 4 of job array 1237
293
294              $ scancel 1237_4
295
296

COPYING

298       Copyright (C) 2002-2007 The Regents of the  University  of  California.
299       Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
300       Copyright (C) 2008-2011 Lawrence Livermore National Security.
301       Copyright (C) 2010-2021 SchedMD LLC.
302
303       This  file  is  part  of Slurm, a resource management program.  For de‐
304       tails, see <https://slurm.schedmd.com/>.
305
306       Slurm is free software; you can redistribute it and/or modify it  under
307       the  terms  of  the GNU General Public License as published by the Free
308       Software Foundation; either version 2 of the License, or (at  your  op‐
309       tion) any later version.
310
311       Slurm  is  distributed  in the hope that it will be useful, but WITHOUT
312       ANY WARRANTY; without even the implied warranty of  MERCHANTABILITY  or
313       FITNESS  FOR  A PARTICULAR PURPOSE.  See the GNU General Public License
314       for more details.
315
316

SEE ALSO

318       slurm_kill_job (3), slurm_kill_job_step (3)
319
320
321
322October 2021                    Slurm Commands                      scancel(1)
Impressum