1SYSTEMD-RUN(1) systemd-run SYSTEMD-RUN(1)
2
6 systemd-run - Run programs in transient scope units, service units, or
7 path-, socket-, or timer-triggered service units
8
10 systemd-run [OPTIONS...] COMMAND [ARGS...]
11 systemd-run [OPTIONS...] [PATH OPTIONS...] {COMMAND} [ARGS...]
13 systemd-run [OPTIONS...] [SOCKET OPTIONS...] {COMMAND} [ARGS...]
15 systemd-run [OPTIONS...] [TIMER OPTIONS...] {COMMAND} [ARGS...]
17
19 systemd-run may be used to create and start a transient .service or
20 .scope unit and run the specified COMMAND in it. It may also be used to
21 create and start a transient .path, .socket, or .timer unit, that
22 activates a .service unit when elapsing.
23 If a command is run as transient service unit, it will be started and
25 managed by the service manager like any other service, and thus shows
26 up in the output of systemctl list-units like any other unit. It will
27 run in a clean and detached execution environment, with the service
28 manager as its parent process. In this mode, systemd-run will start the
29 service asynchronously in the background and return after the command
30 has begun execution (unless --no-block or --wait are specified, see
31 below).
32 If a command is run as transient scope unit, it will be executed by
34 systemd-run itself as parent process and will thus inherit the
35 execution environment of the caller. However, the processes of the
36 command are managed by the service manager similar to normal services,
37 and will show up in the output of systemctl list-units. Execution in
38 this case is synchronous, and will return only when the command
39 finishes. This mode is enabled via the --scope switch (see below).
40 If a command is run with path, socket, or timer options such as
42 --on-calendar= (see below), a transient path, socket, or timer unit is
43 created alongside the service unit for the specified command. Only the
44 transient path, socket, or timer unit is started immediately, the
45 transient service unit will be triggered by the path, socket, or timer
46 unit. If the --unit= option is specified, the COMMAND may be omitted.
47 In this case, systemd-run creates only a .path, .socket, or .timer unit
48 that triggers the specified unit.
49 By default, services created with systemd-run default to the simple
51 type, see the description of Type= in systemd.service(5) for details.
52 Note that when this type is used the service manager (and thus the
53 systemd-run command) considers service start-up successful as soon as
54 the fork() for the main service process succeeded, i.e. before the
55 execve() is invoked, and thus even if the specified command cannot be
56 started. Consider using the exec service type (i.e.
57 --property=Type=exec) to ensure that systemd-run returns successfully
58 only if the specified command line has been successfully started.
59
61 The following options are understood:
62 --no-ask-password
64 Do not query the user for authentication for privileged operations.
65 --scope
67 Create a transient .scope unit instead of the default transient
68 .service unit (see above).
69 --unit=, -u
71 Use this unit name instead of an automatically generated one.
72 --property=, -p
74 Sets a property on the scope or service unit that is created. This
75 option takes an assignment in the same format as systemctl(1)'s
76 set-property command.
77 --description=
79 Provide a description for the service, scope, path, socket, or
80 timer unit. If not specified, the command itself will be used as a
81 description. See Description= in systemd.unit(5).
82 --slice=
84 Make the new .service or .scope unit part of the specified slice,
85 instead of system.slice (when running in --system mode) or the root
86 slice (when running in --user mode).
87 --slice-inherit
89 Make the new .service or .scope unit part of the inherited slice.
90 This option can be combined with --slice=.
91 An inherited slice is located within systemd-run slice. Example: if
93 systemd-run slice is foo.slice, and the --slice= argument is bar,
94 the unit will be placed under the foo-bar.slice.
95 -r, --remain-after-exit
97 After the service process has terminated, keep the service around
98 until it is explicitly stopped. This is useful to collect runtime
99 information about the service after it finished running. Also see
100 RemainAfterExit= in systemd.service(5).
101 --send-sighup
103 When terminating the scope or service unit, send a SIGHUP
104 immediately after SIGTERM. This is useful to indicate to shells and
105 shell-like processes that the connection has been severed. Also see
106 SendSIGHUP= in systemd.kill(5).
107 --service-type=
109 Sets the service type. Also see Type= in systemd.service(5). This
110 option has no effect in conjunction with --scope. Defaults to
111 simple.
112 --uid=, --gid=
114 Runs the service process under the specified UNIX user and group.
115 Also see User= and Group= in systemd.exec(5).
116 --nice=
118 Runs the service process with the specified nice level. Also see
119 Nice= in systemd.exec(5).
120 --working-directory=
122 Runs the service process with the specified working directory. Also
123 see WorkingDirectory= in systemd.exec(5).
124 --same-dir, -d
126 Similar to --working-directory= but uses the current working
127 directory of the caller for the service to execute.
128 -E NAME=VALUE, --setenv=NAME=VALUE
130 Runs the service process with the specified environment variable
131 set. Also see Environment= in systemd.exec(5).
132 --pty, -t
134 When invoking the command, the transient service connects its
135 standard input, output and error to the terminal systemd-run is
136 invoked on, via a pseudo TTY device. This allows running programs
137 that expect interactive user input/output as services, such as
138 interactive command shells.
139 Note that machinectl(1)'s shell command is usually a better
141 alternative for requesting a new, interactive login session on the
142 local host or a local container.
143 See below for details on how this switch combines with --pipe.
145 --pipe, -P
147 If specified, standard input, output, and error of the transient
148 service are inherited from the systemd-run command itself. This
149 allows systemd-run to be used within shell pipelines. Note that
150 this mode is not suitable for interactive command shells and
151 similar, as the service process will not become a TTY controller
152 when invoked on a terminal. Use --pty instead in that case.
153 When both --pipe and --pty are used in combination the more
155 appropriate option is automatically determined and used.
156 Specifically, when invoked with standard input, output and error
157 connected to a TTY --pty is used, and otherwise --pipe.
158 When this option is used the original file descriptors systemd-run
160 receives are passed to the service processes as-is. If the service
161 runs with different privileges than systemd-run, this means the
162 service might not be able to re-open the passed file descriptors,
163 due to normal file descriptor access restrictions. If the invoked
164 process is a shell script that uses the echo "hello" > /dev/stderr
165 construct for writing messages to stderr, this might cause
166 problems, as this only works if stderr can be re-opened. To
167 mitigate this use the construct echo "hello" >&2 instead, which is
168 mostly equivalent and avoids this pitfall.
169 --shell, -S
171 A shortcut for "--pty --same-dir --wait --collect
172 --service-type=exec $SHELL", i.e. requests an interactive shell in
173 the current working directory, running in service context,
174 accessible with a single switch.
175 --quiet, -q
177 Suppresses additional informational output while running. This is
178 particularly useful in combination with --pty when it will suppress
179 the initial message explaining how to terminate the TTY connection.
180 --on-active=, --on-boot=, --on-startup=, --on-unit-active=,
182 --on-unit-inactive=
183 Defines a monotonic timer relative to different starting points for
184 starting the specified command. See OnActiveSec=, OnBootSec=,
185 OnStartupSec=, OnUnitActiveSec= and OnUnitInactiveSec= in
186 systemd.timer(5) for details. These options are shortcuts for
187 --timer-property= with the relevant properties. These options may
188 not be combined with --scope or --pty.
189 --on-calendar=
191 Defines a calendar timer for starting the specified command. See
192 OnCalendar= in systemd.timer(5). This option is a shortcut for
193 --timer-property=OnCalendar=. This option may not be combined with
194 --scope or --pty.
195 --on-clock-change, --on-timezone-change
197 Defines a trigger based on system clock jumps or timezone changes
198 for starting the specified command. See OnClockChange= and
199 OnTimezoneChange= in systemd.timer(5). These options are shortcuts
200 for --timer-property=OnClockChange=yes and
201 --timer-property=OnTimezoneChange=yes. These options may not be
202 combined with --scope or --pty.
203 --path-property=, --socket-property=, --timer-property=
205 Sets a property on the path, socket, or timer unit that is created.
206 This option is similar to --property= but applies to the transient
207 path, socket, or timer unit rather than the transient service unit
208 created. This option takes an assignment in the same format as
209 systemctl(1)'s set-property command. These options may not be
210 combined with --scope or --pty.
211 --no-block
213 Do not synchronously wait for the unit start operation to finish.
214 If this option is not specified, the start request for the
215 transient unit will be verified, enqueued and systemd-run will wait
216 until the unit's start-up is completed. By passing this argument,
217 it is only verified and enqueued. This option may not be combined
218 with --wait.
219 --wait
221 Synchronously wait for the transient service to terminate. If this
222 option is specified, the start request for the transient unit is
223 verified, enqueued, and waited for. Subsequently the invoked unit
224 is monitored, and it is waited until it is deactivated again (most
225 likely because the specified command completed). On exit, terse
226 information about the unit's runtime is shown, including total
227 runtime (as well as CPU usage, if --property=CPUAccounting=1 was
228 set) and the exit code and status of the main process. This output
229 may be suppressed with --quiet. This option may not be combined
230 with --no-block, --scope or the various path, socket, or timer
231 options.
232 -G, --collect
234 Unload the transient unit after it completed, even if it failed.
235 Normally, without this option, all units that ran and failed are
236 kept in memory until the user explicitly resets their failure state
237 with systemctl reset-failed or an equivalent command. On the other
238 hand, units that ran successfully are unloaded immediately. If this
239 option is turned on the "garbage collection" of units is more
240 aggressive, and unloads units regardless if they exited
241 successfully or failed. This option is a shortcut for
242 --property=CollectMode=inactive-or-failed, see the explanation for
243 CollectMode= in systemd.unit(5) for further information.
244 --user
246 Talk to the service manager of the calling user, rather than the
247 service manager of the system.
248 --system
250 Talk to the service manager of the system. This is the implied
251 default.
252 -H, --host=
254 Execute the operation remotely. Specify a hostname, or a username
255 and hostname separated by "@", to connect to. The hostname may
256 optionally be suffixed by a port ssh is listening on, separated by
257 ":", and then a container name, separated by "/", which connects
258 directly to a specific container on the specified host. This will
259 use SSH to talk to the remote machine manager instance. Container
260 names may be enumerated with machinectl -H HOST. Put IPv6 addresses
261 in brackets.
262 -M, --machine=
264 Execute operation on a local container. Specify a container name to
265 connect to, optionally prefixed by a user name to connect as and a
266 separating "@" character. If the special string ".host" is used in
267 place of the container name, a connection to the local system is
268 made (which is useful to connect to a specific user's user bus:
269 "--user --machine=lennart@.host"). If the "@" syntax is not used,
270 the connection is made as root user. If the "@" syntax is used
271 either the left hand side or the right hand side may be omitted
272 (but not both) in which case the local user name and ".host" are
273 implied.
274 -h, --help
276 Print a short help text and exit.
277 --version
279 Print a short version string and exit.
280 All command line arguments after the first non-option argument become
282 part of the command line of the launched process. If a command is run
283 as service unit, the first argument needs to be an absolute program
284 path.
285
287 On success, 0 is returned. If systemd-run failed to start the service,
288 a non-zero return value will be returned. If systemd-run waits for the
289 service to terminate, the return value will be propagated from the
290 service. 0 will be returned on success, including all the cases where
291 systemd considers a service to have exited cleanly, see the discussion
292 of SuccessExitStatus= in systemd.service(5).
293
295 Example 1. Logging environment variables provided by systemd to
296 services
297 # systemd-run env
299 Running as unit: run-19945.service
300 # journalctl -u run-19945.service
301 Sep 08 07:37:21 bupkis systemd[1]: Starting /usr/bin/env...
302 Sep 08 07:37:21 bupkis systemd[1]: Started /usr/bin/env.
303 Sep 08 07:37:21 bupkis env[19948]: PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin
304 Sep 08 07:37:21 bupkis env[19948]: LANG=en_US.UTF-8
305 Sep 08 07:37:21 bupkis env[19948]: BOOT_IMAGE=/vmlinuz-3.11.0-0.rc5.git6.2.fc20.x86_64
306 Example 2. Limiting resources available to a command
308 # systemd-run -p BlockIOWeight=10 updatedb
310 This command invokes the updatedb(8) tool, but lowers the block I/O
312 weight for it to 10. See systemd.resource-control(5) for more
313 information on the BlockIOWeight= property.
314 Example 3. Running commands at a specified time
316 The following command will touch a file after 30 seconds.
318 # date; systemd-run --on-active=30 --timer-property=AccuracySec=100ms /bin/touch /tmp/foo
320 Mon Dec 8 20:44:24 KST 2014
321 Running as unit: run-71.timer
322 Will run service as unit: run-71.service
323 # journalctl -b -u run-71.timer
324 -- Journal begins at Fri 2014-12-05 19:09:21 KST, ends at Mon 2014-12-08 20:44:54 KST. --
325 Dec 08 20:44:38 container systemd[1]: Starting /bin/touch /tmp/foo.
326 Dec 08 20:44:38 container systemd[1]: Started /bin/touch /tmp/foo.
327 # journalctl -b -u run-71.service
328 -- Journal begins at Fri 2014-12-05 19:09:21 KST, ends at Mon 2014-12-08 20:44:54 KST. --
329 Dec 08 20:44:48 container systemd[1]: Starting /bin/touch /tmp/foo...
330 Dec 08 20:44:48 container systemd[1]: Started /bin/touch /tmp/foo.
331 Example 4. Allowing access to the tty
333 The following command invokes /bin/bash as a service passing its
335 standard input, output and error to the calling TTY.
336 # systemd-run -t --send-sighup /bin/bash
338 Example 5. Start screen as a user service
340 $ systemd-run --scope --user screen
342 Running scope as unit run-r14b0047ab6df45bfb45e7786cc839e76.scope.
343 $ screen -ls
345 There is a screen on:
346 492..laptop (Detached)
347 1 Socket in /var/run/screen/S-fatima.
348 This starts the screen process as a child of the systemd --user process
350 that was started by user@.service, in a scope unit. A systemd.scope(5)
351 unit is used instead of a systemd.service(5) unit, because screen will
352 exit when detaching from the terminal, and a service unit would be
353 terminated. Running screen as a user unit has the advantage that it is
354 not part of the session scope. If KillUserProcesses=yes is configured
355 in logind.conf(5), the default, the session scope will be terminated
356 when the user logs out of that session.
357 The user@.service is started automatically when the user first logs in,
359 and stays around as long as at least one login session is open. After
360 the user logs out of the last session, user@.service and all services
361 underneath it are terminated. This behavior is the default, when
362 "lingering" is not enabled for that user. Enabling lingering means that
363 user@.service is started automatically during boot, even if the user is
364 not logged in, and that the service is not terminated when the user
365 logs out.
366 Enabling lingering allows the user to run processes without being
368 logged in, for example to allow screen to persist after the user logs
369 out, even if the session scope is terminated. In the default
370 configuration, users can enable lingering for themselves:
371 $ loginctl enable-linger
373 Example 6. Return value
375 $ systemd-run --user --wait true
377 $ systemd-run --user --wait -p SuccessExitStatus=11 bash -c 'exit 11'
378 $ systemd-run --user --wait -p SuccessExitStatus=SIGUSR1 bash -c 'kill -SIGUSR1 $$$$'
379 Those three invocations will succeed, i.e. terminate with an exit code
381 of 0.
382
384 systemd(1), systemctl(1), systemd.unit(5), systemd.service(5),
385 systemd.scope(5), systemd.slice(5), systemd.exec(5), systemd.resource-
386 control(5), systemd.timer(5), systemd-mount(1), machinectl(1)
387systemd 248 SYSTEMD-RUN(1)