]> git.ipfire.org Git - thirdparty/systemd.git/blob - man/systemd.service.xml
journald: Log error when failed to get machine-id on start
[thirdparty/systemd.git] / man / systemd.service.xml
1 <?xml version='1.0'?> <!--*-nxml-*-->
2 <?xml-stylesheet type="text/xsl" href="http://docbook.sourceforge.net/release/xsl/current/xhtml/docbook.xsl"?>
3 <!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
4 "http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
5
6 <!--
7 This file is part of systemd.
8
9 Copyright 2010 Lennart Poettering
10
11 systemd is free software; you can redistribute it and/or modify it
12 under the terms of the GNU Lesser General Public License as published by
13 the Free Software Foundation; either version 2.1 of the License, or
14 (at your option) any later version.
15
16 systemd is distributed in the hope that it will be useful, but
17 WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 Lesser General Public License for more details.
20
21 You should have received a copy of the GNU Lesser General Public License
22 along with systemd; If not, see <http://www.gnu.org/licenses/>.
23 -->
24
25 <refentry id="systemd.service">
26 <refentryinfo>
27 <title>systemd.service</title>
28 <productname>systemd</productname>
29
30 <authorgroup>
31 <author>
32 <contrib>Developer</contrib>
33 <firstname>Lennart</firstname>
34 <surname>Poettering</surname>
35 <email>lennart@poettering.net</email>
36 </author>
37 </authorgroup>
38 </refentryinfo>
39
40 <refmeta>
41 <refentrytitle>systemd.service</refentrytitle>
42 <manvolnum>5</manvolnum>
43 </refmeta>
44
45 <refnamediv>
46 <refname>systemd.service</refname>
47 <refpurpose>Service unit configuration</refpurpose>
48 </refnamediv>
49
50 <refsynopsisdiv>
51 <para><filename><replaceable>service</replaceable>.service</filename></para>
52 </refsynopsisdiv>
53
54 <refsect1>
55 <title>Description</title>
56
57 <para>A unit configuration file whose name ends in
58 <filename>.service</filename> encodes information
59 about a process controlled and supervised by
60 systemd.</para>
61
62 <para>This man page lists the configuration options
63 specific to this unit type. See
64 <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>
65 for the common options of all unit configuration
66 files. The common configuration items are configured
67 in the generic <literal>[Unit]</literal> and
68 <literal>[Install]</literal> sections. The service
69 specific configuration options are configured in the
70 <literal>[Service]</literal> section.</para>
71
72 <para>Additional options are listed in
73 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
74 which define the execution environment the commands
75 are executed in, and in
76 <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
77 which define the way the processes of the service are
78 terminated, and in
79 <citerefentry><refentrytitle>systemd.cgroup</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
80 which configure control group settings for the
81 processes of the service.</para>
82
83 <para>Unless <varname>DefaultDependencies=</varname>
84 is set to <option>false</option>, service units will
85 implicitly have dependencies of type
86 <varname>Requires=</varname> and
87 <varname>After=</varname> on
88 <filename>basic.target</filename> as well as
89 dependencies of type <varname>Conflicts=</varname> and
90 <varname>Before=</varname> on
91 <filename>shutdown.target</filename>. These ensure
92 that normal service units pull in basic system
93 initialization, and are terminated cleanly prior to
94 system shutdown. Only services involved with early
95 boot or late system shutdown should disable this
96 option.</para>
97
98 <para>If a service is requested under a certain name
99 but no unit configuration file is found, systemd looks
100 for a SysV init script by the same name (with the
101 <filename>.service</filename> suffix removed) and
102 dynamically creates a service unit from that
103 script. This is useful for compatibility with
104 SysV. Note that this compatibility is quite
105 comprehensive but not 100%. For details about the
106 incompatibilities see the <ulink
107 url="http://www.freedesktop.org/wiki/Software/systemd/Incompatibilities">Incompatibilities
108 with SysV</ulink> document.
109 </para>
110 </refsect1>
111
112 <refsect1>
113 <title>Options</title>
114
115 <para>Service files must include a
116 <literal>[Service]</literal> section, which carries
117 information about the service and the process it
118 supervises. A number of options that may be used in
119 this section are shared with other unit types. These
120 options are documented in
121 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
122 and
123 <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>. The
124 options specific to the <literal>[Service]</literal>
125 section of service units are the following:</para>
126
127 <variablelist class='unit-directives'>
128 <varlistentry>
129 <term><varname>Type=</varname></term>
130
131 <listitem><para>Configures the process
132 start-up type for this service
133 unit. One of <option>simple</option>,
134 <option>forking</option>,
135 <option>oneshot</option>,
136 <option>dbus</option>,
137 <option>notify</option> or
138 <option>idle</option>.</para>
139
140 <para>If set to
141 <option>simple</option> (the default
142 value if <varname>BusName=</varname>
143 is not specified) it is expected that
144 the process configured with
145 <varname>ExecStart=</varname> is the
146 main process of the service. In this
147 mode, if the process offers
148 functionality to other processes on
149 the system its communication channels
150 should be installed before the daemon
151 is started up (e.g. sockets set up by
152 systemd, via socket activation), as
153 systemd will immediately proceed
154 starting follow-up units.</para>
155
156 <para>If set to
157 <option>forking</option> it is
158 expected that the process configured
159 with <varname>ExecStart=</varname>
160 will call <function>fork()</function>
161 as part of its start-up. The parent process is
162 expected to exit when start-up is
163 complete and all communication
164 channels set up. The child continues
165 to run as the main daemon
166 process. This is the behavior of
167 traditional UNIX daemons. If this
168 setting is used, it is recommended to
169 also use the
170 <varname>PIDFile=</varname> option, so
171 that systemd can identify the main
172 process of the daemon. systemd will
173 proceed starting follow-up units as
174 soon as the parent process
175 exits.</para>
176
177 <para>Behavior of
178 <option>oneshot</option> is similar
179 to <option>simple</option>, however
180 it is expected that the process has to
181 exit before systemd starts follow-up
182 units. <varname>RemainAfterExit=</varname>
183 is particularly useful for this type
184 of service.</para>
185
186 <para>Behavior of
187 <option>dbus</option> is similar to
188 <option>simple</option>, however it is
189 expected that the daemon acquires a
190 name on the D-Bus bus, as configured
191 by
192 <varname>BusName=</varname>. systemd
193 will proceed starting follow-up units
194 after the D-Bus bus name has been
195 acquired. Service units with this
196 option configured implicitly gain
197 dependencies on the
198 <filename>dbus.socket</filename>
199 unit. This type is the default if
200 <varname>BusName=</varname> is
201 specified.</para>
202
203 <para>Behavior of
204 <option>notify</option> is similar to
205 <option>simple</option>, however it is
206 expected that the daemon sends a
207 notification message via
208 <citerefentry><refentrytitle>sd_notify</refentrytitle><manvolnum>3</manvolnum></citerefentry>
209 or an equivalent call when it finished
210 starting up. systemd will proceed
211 starting follow-up units after this
212 notification message has been sent. If
213 this option is used
214 <varname>NotifyAccess=</varname> (see
215 below) should be set to open access to
216 the notification socket provided by
217 systemd. If
218 <varname>NotifyAccess=</varname> is
219 not set, it will be implicitly set to
220 <option>main</option>. Note that
221 currently
222 <varname>Type=</varname><option>notify</option>
223 won't work if used in combination with
224 <varname>PrivateNetwork=</varname><option>yes</option>.</para>
225
226 <para>Behavior of
227 <option>idle</option> is very similar
228 to <option>simple</option>, however
229 actual execution of the service
230 binary is delayed until all jobs are
231 dispatched. This may be used to avoid
232 interleaving of output of shell
233 services with the status output on the
234 console.</para>
235 </listitem>
236 </varlistentry>
237
238 <varlistentry>
239 <term><varname>RemainAfterExit=</varname></term>
240
241 <listitem><para>Takes a boolean value
242 that specifies whether the service
243 shall be considered active even when
244 all its processes exited. Defaults to
245 <option>no</option>.</para>
246 </listitem>
247 </varlistentry>
248
249 <varlistentry>
250 <term><varname>GuessMainPID=</varname></term>
251
252 <listitem><para>Takes a boolean value
253 that specifies whether systemd should
254 try to guess the main PID of a service
255 if it cannot be determined
256 reliably. This option is ignored
257 unless <option>Type=forking</option>
258 is set and <option>PIDFile=</option>
259 is unset because for the other types
260 or with an explicitly configured PID
261 file the main PID is always known. The
262 guessing algorithm might come to
263 incorrect conclusions if a daemon
264 consists of more than one process. If
265 the main PID cannot be determined
266 failure detection and automatic
267 restarting of a service will not work
268 reliably. Defaults to
269 <option>yes</option>.</para>
270 </listitem>
271 </varlistentry>
272
273 <varlistentry>
274 <term><varname>PIDFile=</varname></term>
275
276 <listitem><para>Takes an absolute file
277 name pointing to the PID file of this
278 daemon. Use of this option is
279 recommended for services where
280 <varname>Type=</varname> is set to
281 <option>forking</option>. systemd will
282 read the PID of the main process of
283 the daemon after start-up of the
284 service. systemd will not write to the
285 file configured here.</para>
286 </listitem>
287 </varlistentry>
288
289 <varlistentry>
290 <term><varname>BusName=</varname></term>
291
292 <listitem><para>Takes a D-Bus bus
293 name, that this service is reachable
294 as. This option is mandatory for
295 services where
296 <varname>Type=</varname> is set to
297 <option>dbus</option>, but its use
298 is otherwise recommended as well if
299 the process takes a name on the D-Bus
300 bus.</para>
301 </listitem>
302 </varlistentry>
303
304 <varlistentry>
305 <term><varname>ExecStart=</varname></term>
306 <listitem><para>Commands with their
307 arguments that are executed when this
308 service is started. The first
309 argument must be an absolute path
310 name.</para>
311
312 <para>When <varname>Type</varname> is
313 not <option>oneshot</option>, only one
314 command may be given. When
315 <varname>Type=oneshot</varname> is
316 used, more than one command may be
317 specified. Multiple command lines may
318 be concatenated in a single directive,
319 by separating them with semicolons
320 (these semicolons must be passed as
321 separate words). Alternatively, this
322 directive may be specified more than
323 once with the same effect. However,
324 the latter syntax is not recommended
325 for compatibility with parsers
326 suitable for XDG
327 <filename>.desktop</filename> files.
328 Lone semicolons may be escaped as
329 <literal>\;</literal>. If the empty
330 string is assigned to this option the
331 list of commands to start is reset,
332 prior assignments of this option will
333 have no effect.</para>
334
335 <para>If more than one command is
336 specified, the commands are invoked
337 one by one sequentially in the order
338 they appear in the unit file. If one
339 of the commands fails (and is not
340 prefixed with <literal>-</literal>),
341 other lines are not executed and the
342 unit is considered failed.</para>
343
344 <para>Unless
345 <varname>Type=forking</varname> is
346 set, the process started via this
347 command line will be considered the
348 main process of the daemon.</para>
349
350 <para>The command line accepts
351 <literal>%</literal> specifiers as
352 described in
353 <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>. Note
354 that the first argument of the command
355 line (i.e. the program to execute) may
356 not include specifiers.</para>
357
358 <para>Basic environment variable
359 substitution is supported. Use
360 <literal>${FOO}</literal> as part of a
361 word, or as a word of its own on the
362 command line, in which case it will be
363 replaced by the value of the
364 environment variable including all
365 whitespace it contains, resulting in a
366 single argument. Use
367 <literal>$FOO</literal> as a separate
368 word on the command line, in which
369 case it will be replaced by the value
370 of the environment variable split up
371 at whitespace, resulting in zero or
372 more arguments. Note that the first
373 argument (i.e. the program to execute)
374 may not be a variable, since it must
375 be a literal and absolute path
376 name.</para>
377
378 <para>Optionally, if the absolute file
379 name is prefixed with
380 <literal>@</literal>, the second token
381 will be passed as
382 <literal>argv[0]</literal> to the
383 executed process, followed by the
384 further arguments specified. If the
385 absolute filename is prefixed with
386 <literal>-</literal> an exit code of
387 the command normally considered a
388 failure (i.e. non-zero exit status or
389 abnormal exit due to signal) is ignored
390 and considered success. If both
391 <literal>-</literal> and
392 <literal>@</literal> are used they
393 can appear in either order.</para>
394
395 <para>Note that this setting does not
396 directly support shell command
397 lines. If shell command lines are to
398 be used they need to be passed
399 explicitly to a shell implementation
400 of some kind. Example:</para>
401 <programlisting>ExecStart=/bin/sh -c 'dmesg | tac'
402 </programlisting>
403
404 <para>For services run by a user
405 instance of systemd the special
406 environment variable
407 <varname>$MANAGERPID</varname> is set
408 to the PID of the systemd
409 instance.</para>
410 </listitem>
411 </varlistentry>
412
413 <varlistentry>
414 <term><varname>ExecStartPre=</varname></term>
415 <term><varname>ExecStartPost=</varname></term>
416 <listitem><para>Additional commands
417 that are executed before or after
418 the command in
419 <varname>ExecStart=</varname>, respectively.
420 Syntax is the same as for
421 <varname>ExecStart=</varname>, except
422 that multiple command lines are allowed
423 and the commands are executed one
424 after the other, serially.</para>
425
426 <para>If any of those commands (not
427 prefixed with <literal>-</literal>)
428 fail, the rest are not executed and
429 the unit is considered failed.</para>
430 </listitem>
431 </varlistentry>
432
433 <varlistentry>
434 <term><varname>ExecReload=</varname></term>
435 <listitem><para>Commands to execute to
436 trigger a configuration reload in the
437 service. This argument takes multiple
438 command lines, following the same
439 scheme as described for
440 <varname>ExecStart=</varname>
441 above. Use of this setting is
442 optional. Specifier and environment
443 variable substitution is supported
444 here following the same scheme as for
445 <varname>ExecStart=</varname>.</para>
446
447 <para>One additional special
448 environment variables is set: if known
449 <varname>$MAINPID</varname> is set to
450 the main process of the daemon, and
451 may be used for command lines like the
452 following:</para>
453
454 <programlisting>/bin/kill -HUP $MAINPID</programlisting>
455 </listitem>
456 </varlistentry>
457
458 <varlistentry>
459 <term><varname>ExecStop=</varname></term>
460 <listitem><para>Commands to execute to
461 stop the service started via
462 <varname>ExecStart=</varname>. This
463 argument takes multiple command lines,
464 following the same scheme as described
465 for <varname>ExecStart=</varname>
466 above. Use of this setting is
467 optional. All processes remaining for
468 a service after the commands
469 configured in this option are run are
470 terminated according to the
471 <varname>KillMode=</varname> setting
472 (see
473 <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>). If
474 this option is not specified the
475 process is terminated right-away when
476 service stop is requested. Specifier
477 and environment variable substitution
478 is supported (including
479 <varname>$MAINPID</varname>, see
480 above).</para></listitem>
481 </varlistentry>
482
483 <varlistentry>
484 <term><varname>ExecStopPost=</varname></term>
485 <listitem><para>Additional commands
486 that are executed after the service
487 was stopped. This includes cases where
488 the commands configured in
489 <varname>ExecStop=</varname> were used,
490 where the service does not have any
491 <varname>ExecStop=</varname> defined, or
492 where the service exited unexpectedly. This
493 argument takes multiple command lines,
494 following the same scheme as described
495 for <varname>ExecStart</varname>. Use
496 of these settings is
497 optional. Specifier and environment
498 variable substitution is
499 supported.</para></listitem>
500 </varlistentry>
501
502 <varlistentry>
503 <term><varname>RestartSec=</varname></term>
504 <listitem><para>Configures the time to
505 sleep before restarting a service (as
506 configured with
507 <varname>Restart=</varname>). Takes a
508 unit-less value in seconds, or a time
509 span value such as "5min
510 20s". Defaults to
511 100ms.</para></listitem>
512 </varlistentry>
513
514 <varlistentry>
515 <term><varname>TimeoutStartSec=</varname></term>
516 <listitem><para>Configures the time to
517 wait for start-up. If a
518 daemon service does not signal
519 start-up completion within the
520 configured time, the service will be
521 considered failed and be shut down
522 again.
523 Takes a unit-less value in seconds, or a
524 time span value such as "5min
525 20s". Pass 0 to disable the timeout
526 logic. Defaults to 90s, except when
527 <varname>Type=oneshot</varname> is
528 used in which case the timeout
529 is disabled by default.
530 </para></listitem>
531 </varlistentry>
532
533 <varlistentry>
534 <term><varname>TimeoutStopSec=</varname></term>
535 <listitem><para>Configures the time to
536 wait for stop. If a service is asked
537 to stop but does not terminate in the
538 specified time, it will be terminated
539 forcibly via <constant>SIGTERM</constant>, and after
540 another delay of this time with
541 <constant>SIGKILL</constant> (See
542 <varname>KillMode=</varname>
543 in <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>).
544 Takes a unit-less value in seconds, or a
545 time span value such as "5min
546 20s". Pass 0 to disable the timeout
547 logic. Defaults to 90s.
548 </para></listitem>
549 </varlistentry>
550
551 <varlistentry>
552 <term><varname>TimeoutSec=</varname></term>
553 <listitem><para>A shorthand for configuring
554 both <varname>TimeoutStartSec=</varname>
555 and <varname>TimeoutStopSec=</varname>
556 to the specified value.
557 </para></listitem>
558 </varlistentry>
559
560 <varlistentry>
561 <term><varname>WatchdogSec=</varname></term>
562 <listitem><para>Configures the
563 watchdog timeout for a service. The
564 watchdog is activated when the start-up is
565 completed. The service must call
566 <citerefentry><refentrytitle>sd_notify</refentrytitle><manvolnum>3</manvolnum></citerefentry>
567 regularly with "WATCHDOG=1" (i.e. the
568 "keep-alive ping"). If the time
569 between two such calls is larger than
570 the configured time then the service
571 is placed in a failure state. By
572 setting <varname>Restart=</varname> to
573 <option>on-failure</option> or
574 <option>always</option> the service
575 will be automatically restarted. The
576 time configured here will be passed to
577 the executed service process in the
578 <varname>WATCHDOG_USEC=</varname>
579 environment variable. This allows
580 daemons to automatically enable the
581 keep-alive pinging logic if watchdog
582 support is enabled for the service. If
583 this option is used
584 <varname>NotifyAccess=</varname> (see
585 below) should be set to open access to
586 the notification socket provided by
587 systemd. If
588 <varname>NotifyAccess=</varname> is
589 not set, it will be implicitly set to
590 <option>main</option>. Defaults to 0,
591 which disables this
592 feature.</para></listitem>
593 </varlistentry>
594
595 <varlistentry>
596 <term><varname>Restart=</varname></term>
597 <listitem><para>Configures whether the
598 service shall be restarted when the
599 service process exits, is killed,
600 or a timeout is reached. The service
601 process may be the main service
602 process, but also one of the processes
603 specified with
604 <varname>ExecStartPre=</varname>,
605 <varname>ExecStartPost=</varname>,
606 <varname>ExecStopPre=</varname>,
607 <varname>ExecStopPost=</varname>, or
608 <varname>ExecReload=</varname>.
609 When the death of the process is a
610 result of systemd operation (e.g. service
611 stop or restart), the service will not be
612 restarted. Timeouts include missing
613 the watchdog "keep-alive ping"
614 deadline and a service start, reload,
615 and stop operation timeouts.</para>
616
617 <para>Takes one of
618 <option>no</option>,
619 <option>on-success</option>,
620 <option>on-failure</option>,
621 <option>on-abort</option>, or
622 <option>always</option>. If set to
623 <option>no</option> (the default) the
624 service will not be restarted. If set to
625 <option>on-success</option> it will be
626 restarted only when the service process
627 exits cleanly.
628 In this context, a clean exit means
629 an exit code of 0, or one of the signals
630 <constant>SIGHUP</constant>, <constant>SIGINT</constant>, <constant>SIGTERM</constant>, or <constant>SIGPIPE</constant>, and
631 additionally, exit statuses and signals
632 specified in <varname>SuccessExitStatus=</varname>.
633 If set to <option>on-failure</option>
634 the service will be restarted when the
635 process exits with an nonzero exit code,
636 is terminated by a signal (including on
637 core dump), when an operation (such as
638 service reload) times out, and when the
639 configured watchdog timeout is triggered.
640 If set to
641 <option>on-abort</option> the service
642 will be restarted only if the service
643 process exits due to an uncaught
644 signal not specified as a clean exit
645 status.
646 If set to
647 <option>always</option> the service
648 will be restarted regardless whether
649 it exited cleanly or not, got
650 terminated abnormally by a signal or
651 hit a timeout.</para>
652
653 <para>In addition to the above settings,
654 the service will not be restarted if the
655 exit code or signal is specified in
656 <varname>RestartPreventExitStatus=</varname>
657 (see below).</para></listitem>
658 </varlistentry>
659
660 <varlistentry>
661 <term><varname>SuccessExitStatus=</varname></term>
662 <listitem><para>Takes a list of exit
663 status definitions that when returned
664 by the main service process will be
665 considered successful termination, in
666 addition to the normal successful exit
667 code 0 and the signals <constant>SIGHUP</constant>, <constant>SIGINT</constant>,
668 <constant>SIGTERM</constant> and <constant>SIGPIPE</constant>. Exit status
669 definitions can either be numeric exit
670 codes or termination signal names,
671 separated by spaces. Example:
672 <literal>SuccessExitStatus=1 2 8
673 <constant>SIGKILL</constant></literal>, ensures that exit
674 codes 1, 2, 8 and the termination
675 signal <constant>SIGKILL</constant> are considered clean
676 service terminations. This option may
677 appear more than once in which case
678 the list of successful exit statuses
679 is merged. If the empty string is
680 assigned to this option the list is
681 reset, all prior assignments of this
682 option will have no
683 effect.</para></listitem>
684 </varlistentry>
685
686 <varlistentry>
687 <term><varname>RestartPreventExitStatus=</varname></term>
688 <listitem><para>Takes a list of exit
689 status definitions that when returned
690 by the main service process will
691 prevent automatic service restarts
692 regardless of the restart setting
693 configured with
694 <varname>Restart=</varname>. Exit
695 status definitions can either be
696 numeric exit codes or termination
697 signal names, and are separated by
698 spaces. Defaults to the empty list, so
699 that by default no exit status is
700 excluded from the configured restart
701 logic. Example:
702 <literal>RestartPreventExitStatus=1 6
703 SIGABRT</literal>, ensures that exit
704 codes 1 and 6 and the termination
705 signal SIGABRT will not result in
706 automatic service restarting. This
707 option may appear more than once in
708 which case the list of restart preventing
709 statuses is merged. If the empty
710 string is assigned to this option the
711 list is reset, all prior assignments
712 of this option will have no
713 effect.</para></listitem>
714 </varlistentry>
715
716 <varlistentry>
717 <term><varname>PermissionsStartOnly=</varname></term>
718 <listitem><para>Takes a boolean
719 argument. If true, the permission
720 related execution options as
721 configured with
722 <varname>User=</varname> and similar
723 options (see
724 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
725 for more information) are only applied
726 to the process started with
727 <varname>ExecStart=</varname>, and not
728 to the various other
729 <varname>ExecStartPre=</varname>,
730 <varname>ExecStartPost=</varname>,
731 <varname>ExecReload=</varname>,
732 <varname>ExecStop=</varname>,
733 <varname>ExecStopPost=</varname>
734 commands. If false, the setting is
735 applied to all configured commands the
736 same way. Defaults to
737 false.</para></listitem>
738 </varlistentry>
739
740 <varlistentry>
741 <term><varname>RootDirectoryStartOnly=</varname></term>
742 <listitem><para>Takes a boolean
743 argument. If true, the root directory
744 as configured with the
745 <varname>RootDirectory=</varname>
746 option (see
747 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
748 for more information) is only applied
749 to the process started with
750 <varname>ExecStart=</varname>, and not
751 to the various other
752 <varname>ExecStartPre=</varname>,
753 <varname>ExecStartPost=</varname>,
754 <varname>ExecReload=</varname>,
755 <varname>ExecStop=</varname>,
756 <varname>ExecStopPost=</varname>
757 commands. If false, the setting is
758 applied to all configured commands the
759 same way. Defaults to
760 false.</para></listitem>
761 </varlistentry>
762
763 <varlistentry>
764 <term><varname>NonBlocking=</varname></term>
765 <listitem><para>Set O_NONBLOCK flag
766 for all file descriptors passed via
767 socket-based activation. If true, all
768 file descriptors >= 3 (i.e. all except
769 STDIN/STDOUT/STDERR) will have
770 the O_NONBLOCK flag set and hence are in
771 non-blocking mode. This option is only
772 useful in conjunction with a socket
773 unit, as described in
774 <citerefentry><refentrytitle>systemd.socket</refentrytitle><manvolnum>5</manvolnum></citerefentry>. Defaults
775 to false.</para></listitem>
776 </varlistentry>
777
778 <varlistentry>
779 <term><varname>NotifyAccess=</varname></term>
780 <listitem><para>Controls access to the
781 service status notification socket, as
782 accessible via the
783 <citerefentry><refentrytitle>sd_notify</refentrytitle><manvolnum>3</manvolnum></citerefentry>
784 call. Takes one of
785 <option>none</option> (the default),
786 <option>main</option> or
787 <option>all</option>. If
788 <option>none</option> no daemon status
789 updates are accepted from the service
790 processes, all status update messages
791 are ignored. If <option>main</option>
792 only service updates sent from the
793 main process of the service are
794 accepted. If <option>all</option> all
795 services updates from all members of
796 the service's control group are
797 accepted. This option should be set to
798 open access to the notification socket
799 when using
800 <varname>Type=notify</varname> or
801 <varname>WatchdogSec=</varname> (see
802 above). If those options are used but
803 <varname>NotifyAccess=</varname> not
804 configured it will be implicitly set
805 to
806 <option>main</option>.</para></listitem>
807 </varlistentry>
808
809 <varlistentry>
810 <term><varname>Sockets=</varname></term>
811 <listitem><para>Specifies the name of
812 the socket units this service shall
813 inherit the sockets from when the
814 service is started. Normally it
815 should not be necessary to use this
816 setting as all sockets whose unit
817 shares the same name as the service
818 (ignoring the different suffix of course)
819 are passed to the spawned
820 process.</para>
821
822 <para>Note that the same socket may be
823 passed to multiple processes at the
824 same time. Also note that a different
825 service may be activated on incoming
826 traffic than inherits the sockets. Or
827 in other words: the
828 <varname>Service=</varname> setting of
829 <filename>.socket</filename> units
830 does not have to match the inverse of
831 the <varname>Sockets=</varname>
832 setting of the
833 <filename>.service</filename> it
834 refers to.</para>
835
836 <para>This option may appear more than
837 once, in which case the list of socket
838 units is merged. If the empty string
839 is assigned to this option the list of
840 sockets is reset, all prior uses of
841 this setting will have no
842 effect.</para></listitem>
843 </varlistentry>
844
845 <varlistentry>
846 <term><varname>StartLimitInterval=</varname></term>
847 <term><varname>StartLimitBurst=</varname></term>
848
849 <listitem><para>Configure service
850 start rate limiting. By default
851 services which are started more often
852 than 5 times within 10s are not
853 permitted to start any more times
854 until the 10s interval ends. With
855 these two options this rate limiting
856 may be modified. Use
857 <varname>StartLimitInterval=</varname>
858 to configure the checking interval
859 (defaults to 10s, set to 0 to disable
860 any kind of rate limiting). Use
861 <varname>StartLimitBurst=</varname> to
862 configure how many starts per interval
863 are allowed (defaults to 5). These
864 configuration options are particularly
865 useful in conjunction with
866 <varname>Restart=</varname>, however
867 apply to all kinds of starts
868 (including manual), not just those
869 triggered by the
870 <varname>Restart=</varname> logic.
871 Note that units which are configured
872 for <varname>Restart=</varname> and
873 which reach the start limit are not
874 attempted to be restarted anymore,
875 however they may still be restarted
876 manually at a later point from which
877 point on the restart logic is again
878 activated. Note that
879 <command>systemctl
880 reset-failed</command> will cause the
881 restart rate counter for a service to
882 be flushed, which is useful if the
883 administrator wants to manually start
884 a service and the start limit
885 interferes with
886 that.</para></listitem>
887 </varlistentry>
888
889 <varlistentry>
890 <term><varname>StartLimitAction=</varname></term>
891
892 <listitem><para>Configure the action
893 to take if the rate limit configured
894 with
895 <varname>StartLimitInterval=</varname>
896 and
897 <varname>StartLimitBurst=</varname> is
898 hit. Takes one of
899 <option>none</option>,
900 <option>reboot</option>,
901 <option>reboot-force</option> or
902 <option>reboot-immediate</option>. If
903 <option>none</option> is set,
904 hitting the rate limit will trigger no
905 action besides that the start will not
906 be
907 permitted. <option>reboot</option>
908 causes a reboot following the normal
909 shutdown procedure (i.e. equivalent to
910 <command>systemctl reboot</command>),
911 <option>reboot-force</option> causes
912 an forced reboot which will terminate
913 all processes forcibly but should
914 cause no dirty file systems on reboot
915 (i.e. equivalent to <command>systemctl
916 reboot -f</command>) and
917 <option>reboot-immediate</option>
918 causes immediate execution of the
919 <citerefentry><refentrytitle>reboot</refentrytitle><manvolnum>2</manvolnum></citerefentry>
920 system call, which might result in
921 data loss. Defaults to
922 <option>none</option>.</para></listitem>
923 </varlistentry>
924
925 </variablelist>
926
927 <para>Check
928 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>
929 and
930 <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>
931 for more settings.</para>
932
933 </refsect1>
934
935 <refsect1>
936 <title>Compatibility Options</title>
937
938 <para>The following options are also available in the
939 <literal>[Service]</literal> section, but exist purely
940 for compatibility reasons and should not be used in
941 newly written service files.</para>
942
943 <variablelist class='unit-directives'>
944 <varlistentry>
945 <term><varname>SysVStartPriority=</varname></term>
946 <listitem><para>Set the SysV start
947 priority to use to order this service
948 in relation to SysV services lacking
949 LSB headers. This option is only
950 necessary to fix ordering in relation
951 to legacy SysV services, that have no
952 ordering information encoded in the
953 script headers. As such it should only
954 be used as temporary compatibility
955 option, and not be used in new unit
956 files. Almost always it is a better
957 choice to add explicit ordering
958 directives via
959 <varname>After=</varname> or
960 <varname>Before=</varname>,
961 instead. For more details see
962 <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>. If
963 used, pass an integer value in the
964 range 0-99.</para></listitem>
965 </varlistentry>
966
967 <varlistentry>
968 <term><varname>FsckPassNo=</varname></term>
969 <listitem><para>Set the fsck passno
970 priority to use to order this service
971 in relation to other file system
972 checking services. This option is only
973 necessary to fix ordering in relation
974 to fsck jobs automatically created for
975 all <filename>/etc/fstab</filename>
976 entries with a value in the fs_passno
977 column > 0. As such it should only be
978 used as option for fsck
979 services. Almost always it is a better
980 choice to add explicit ordering
981 directives via
982 <varname>After=</varname> or
983 <varname>Before=</varname>,
984 instead. For more details see
985 <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>. If
986 used, pass an integer value in the
987 same range as
988 <filename>/etc/fstab</filename>'s
989 fs_passno column. See
990 <citerefentry><refentrytitle>fstab</refentrytitle><manvolnum>5</manvolnum></citerefentry>
991 for details.</para></listitem>
992 </varlistentry>
993
994 </variablelist>
995 </refsect1>
996
997 <refsect1>
998 <title>See Also</title>
999 <para>
1000 <citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry>,
1001 <citerefentry><refentrytitle>systemctl</refentrytitle><manvolnum>8</manvolnum></citerefentry>,
1002 <citerefentry><refentrytitle>systemd.unit</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1003 <citerefentry><refentrytitle>systemd.exec</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1004 <citerefentry><refentrytitle>systemd.cgroup</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1005 <citerefentry><refentrytitle>systemd.kill</refentrytitle><manvolnum>5</manvolnum></citerefentry>,
1006 <citerefentry><refentrytitle>systemd.directives</refentrytitle><manvolnum>7</manvolnum></citerefentry>
1007 </para>
1008 </refsect1>
1009
1010 </refentry>