1SYSTEMD-ANALYZE(1) systemd-analyze SYSTEMD-ANALYZE(1)
2
6 systemd-analyze - Analyze and debug system manager
7
9 systemd-analyze [OPTIONS...] [time]
10 systemd-analyze [OPTIONS...] blame
12 systemd-analyze [OPTIONS...] critical-chain [UNIT...]
14 systemd-analyze [OPTIONS...] log-level [LEVEL]
16 systemd-analyze [OPTIONS...] log-target [TARGET]
18 systemd-analyze [OPTIONS...] service-watchdogs [BOOL]
20 systemd-analyze [OPTIONS...] dump
22 systemd-analyze [OPTIONS...] plot [>file.svg]
24 systemd-analyze [OPTIONS...] dot [PATTERN...] [>file.dot]
26 systemd-analyze [OPTIONS...] unit-paths
28 systemd-analyze [OPTIONS...] exit-status [STATUS...]
30 systemd-analyze [OPTIONS...] condition CONDITION...
32 systemd-analyze [OPTIONS...] syscall-filter [SET...]
34 systemd-analyze [OPTIONS...] calendar SPEC...
36 systemd-analyze [OPTIONS...] timestamp TIMESTAMP...
38 systemd-analyze [OPTIONS...] timespan SPAN...
40 systemd-analyze [OPTIONS...] cat-config NAME|PATH...
42 systemd-analyze [OPTIONS...] verify [FILE...]
44 systemd-analyze [OPTIONS...] security UNIT...
46
48 systemd-analyze may be used to determine system boot-up performance
49 statistics and retrieve other state and tracing information from the
50 system and service manager, and to verify the correctness of unit
51 files. It is also used to access special functions useful for advanced
52 system manager debugging.
53 If no command is passed, systemd-analyze time is implied.
55 systemd-analyze time
57 This command prints the time spent in the kernel before userspace has
58 been reached, the time spent in the initial RAM disk (initrd) before
59 normal system userspace has been reached, and the time normal system
60 userspace took to initialize. Note that these measurements simply
61 measure the time passed up to the point where all system services have
62 been spawned, but not necessarily until they fully finished
63 initialization or the disk is idle.
64 Example 1. Show how long the boot took
66 # in a container
68 $ systemd-analyze time
69 Startup finished in 296ms (userspace)
70 multi-user.target reached after 275ms in userspace
71 # on a real machine
73 $ systemd-analyze time
74 Startup finished in 2.584s (kernel) + 19.176s (initrd) + 47.847s (userspace) = 1min 9.608s
75 multi-user.target reached after 47.820s in userspace
76 systemd-analyze blame
78 This command prints a list of all running units, ordered by the time
79 they took to initialize. This information may be used to optimize
80 boot-up times. Note that the output might be misleading as the
81 initialization of one service might be slow simply because it waits for
82 the initialization of another service to complete. Also note:
83 systemd-analyze blame doesn't display results for services with
84 Type=simple, because systemd considers such services to be started
85 immediately, hence no measurement of the initialization delays can be
86 done. Also note that this command only shows the time units took for
87 starting up, it does not show how long unit jobs spent in the execution
88 queue. In particular it shows the time units spent in "activating"
89 state, which is not defined for units such as device units that
90 transition directly from "inactive" to "active". This command hence
91 gives an impression of the performance of program code, but cannot
92 accurately reflect latency introduced by waiting for hardware and
93 similar events.
94 Example 2. Show which units took the most time during boot
96 $ systemd-analyze blame
98 32.875s pmlogger.service
99 20.905s systemd-networkd-wait-online.service
100 13.299s dev-vda1.device
101 ...
102 23ms sysroot.mount
103 11ms initrd-udevadm-cleanup-db.service
104 3ms sys-kernel-config.mount
105 systemd-analyze critical-chain [UNIT...]
108 This command prints a tree of the time-critical chain of units (for
109 each of the specified UNITs or for the default target otherwise). The
110 time after the unit is active or started is printed after the "@"
111 character. The time the unit takes to start is printed after the "+"
112 character. Note that the output might be misleading as the
113 initialization of services might depend on socket activation and
114 because of the parallel execution of units. Also, similar to the blame
115 command, this only takes into account the time units spent in
116 "activating" state, and hence does not cover units that never went
117 through an "activating" state (such as device units that transition
118 directly from "inactive" to "active"). Moreover it does not show
119 information on jobs (and in particular not jobs that timed out).
120 Example 3. systemd-analyze time
122 $ systemd-analyze critical-chain
124 multi-user.target @47.820s
125 └─pmie.service @35.968s +548ms
126 └─pmcd.service @33.715s +2.247s
127 └─network-online.target @33.712s
128 └─systemd-networkd-wait-online.service @12.804s +20.905s
129 └─systemd-networkd.service @11.109s +1.690s
130 └─systemd-udevd.service @9.201s +1.904s
131 └─systemd-tmpfiles-setup-dev.service @7.306s +1.776s
132 └─kmod-static-nodes.service @6.976s +177ms
133 └─systemd-journald.socket
134 └─system.slice
135 └─-.slice
136 systemd-analyze log-level [LEVEL]
138 systemd-analyze log-level prints the current log level of the systemd
139 daemon. If an optional argument LEVEL is provided, then the command
140 changes the current log level of the systemd daemon to LEVEL (accepts
141 the same values as --log-level= described in systemd(1)).
142 systemd-analyze log-target [TARGET]
144 systemd-analyze log-target prints the current log target of the systemd
145 daemon. If an optional argument TARGET is provided, then the command
146 changes the current log target of the systemd daemon to TARGET (accepts
147 the same values as --log-target=, described in systemd(1)).
148 systemd-analyze service-watchdogs [yes|no]
150 systemd-analyze service-watchdogs prints the current state of service
151 runtime watchdogs of the systemd daemon. If an optional boolean
152 argument is provided, then globally enables or disables the service
153 runtime watchdogs (WatchdogSec=) and emergency actions (e.g.
154 OnFailure= or StartLimitAction=); see systemd.service(5). The hardware
155 watchdog is not affected by this setting.
156 systemd-analyze dump
158 This command outputs a (usually very long) human-readable serialization
159 of the complete server state. Its format is subject to change without
160 notice and should not be parsed by applications.
161 Example 4. Show the internal state of user manager
163 $ systemd-analyze --user dump
165 Timestamp userspace: Thu 2019-03-14 23:28:07 CET
166 Timestamp finish: Thu 2019-03-14 23:28:07 CET
167 Timestamp generators-start: Thu 2019-03-14 23:28:07 CET
168 Timestamp generators-finish: Thu 2019-03-14 23:28:07 CET
169 Timestamp units-load-start: Thu 2019-03-14 23:28:07 CET
170 Timestamp units-load-finish: Thu 2019-03-14 23:28:07 CET
171 -> Unit proc-timer_list.mount:
172 Description: /proc/timer_list
173 ...
174 -> Unit default.target:
175 Description: Main user target
176 ...
177 systemd-analyze plot
179 This command prints an SVG graphic detailing which system services have
180 been started at what time, highlighting the time they spent on
181 initialization.
182 Example 5. Plot a bootchart
184 $ systemd-analyze plot >bootup.svg
186 $ eog bootup.svg&
187 systemd-analyze dot [pattern...]
189 This command generates textual dependency graph description in dot
190 format for further processing with the GraphViz dot(1) tool. Use a
191 command line like systemd-analyze dot | dot -Tsvg >systemd.svg to
192 generate a graphical dependency tree. Unless --order or --require is
193 passed, the generated graph will show both ordering and requirement
194 dependencies. Optional pattern globbing style specifications (e.g.
195 *.target) may be given at the end. A unit dependency is included in the
196 graph if any of these patterns match either the origin or destination
197 node.
198 Example 6. Plot all dependencies of any unit whose name starts with
200 "avahi-daemon"
201 $ systemd-analyze dot 'avahi-daemon.*' | dot -Tsvg >avahi.svg
203 $ eog avahi.svg
204 Example 7. Plot the dependencies between all known target units
206 $ systemd-analyze dot --to-pattern='*.target' --from-pattern='*.target' \
208 | dot -Tsvg >targets.svg
209 $ eog targets.svg
210 systemd-analyze unit-paths
212 This command outputs a list of all directories from which unit files,
213 .d overrides, and .wants, .requires symlinks may be loaded. Combine
214 with --user to retrieve the list for the user manager instance, and
215 --global for the global configuration of user manager instances.
216 Example 8. Show all paths for generated units
218 $ systemd-analyze unit-paths | grep '^/run'
220 /run/systemd/system.control
221 /run/systemd/transient
222 /run/systemd/generator.early
223 /run/systemd/system
224 /run/systemd/system.attached
225 /run/systemd/generator
226 /run/systemd/generator.late
227 Note that this verb prints the list that is compiled into
229 systemd-analyze itself, and does not communicate with the running
230 manager. Use
231 systemctl [--user] [--global] show -p UnitPath --value
233 to retrieve the actual list that the manager uses, with any empty
235 directories omitted.
236 systemd-analyze exit-status [STATUS...]
238 This command prints a list of exit statuses along with their "class",
239 i.e. the source of the definition (one of "glibc", "systemd", "LSB", or
240 "BSD"), see the Process Exit Codes section in systemd.exec(5). If no
241 additional arguments are specified, all known statuses are are shown.
242 Otherwise, only the definitions for the specified codes are shown.
243 Example 9. Show some example exit status names
245 $ systemd-analyze exit-status 0 1 {63..65}
247 NAME STATUS CLASS
248 SUCCESS 0 glibc
249 FAILURE 1 glibc
250 - 63 -
251 USAGE 64 BSD
252 DATAERR 65 BSD
253 systemd-analyze condition CONDITION...
255 This command will evaluate Condition*=... and Assert*=...
256 assignments, and print their values, and the resulting value of the
257 combined condition set. See systemd.unit(5) for a list of available
258 conditions and asserts.
259 Example 10. Evaluate conditions that check kernel versions
261 $ systemd-analyze condition 'ConditionKernelVersion = ! <4.0' \
263 'ConditionKernelVersion = >=5.1' \
264 'ConditionACPower=|false' \
265 'ConditionArchitecture=|!arm' \
266 'AssertPathExists=/etc/os-release'
267 test.service: AssertPathExists=/etc/os-release succeeded.
268 Asserts succeeded.
269 test.service: ConditionArchitecture=|!arm succeeded.
270 test.service: ConditionACPower=|false failed.
271 test.service: ConditionKernelVersion=>=5.1 succeeded.
272 test.service: ConditionKernelVersion=!<4.0 succeeded.
273 Conditions succeeded.
274 systemd-analyze syscall-filter [SET...]
276 This command will list system calls contained in the specified system
277 call set SET, or all known sets if no sets are specified. Argument SET
278 must include the "@" prefix.
279 systemd-analyze calendar EXPRESSION...
281 This command will parse and normalize repetitive calendar time events,
282 and will calculate when they elapse next. This takes the same input as
283 the OnCalendar= setting in systemd.timer(5), following the syntax
284 described in systemd.time(7). By default, only the next time the
285 calendar expression will elapse is shown; use --iterations= to show the
286 specified number of next times the expression elapses. Each time the
287 expression elapses forms a timestamp, see the timestamp verb below.
288 Example 11. Show leap days in the near future
290 $ systemd-analyze calendar --iterations=5 '*-2-29 0:0:0'
292 Original form: *-2-29 0:0:0
293 Normalized form: *-02-29 00:00:00
294 Next elapse: Sat 2020-02-29 00:00:00 UTC
295 From now: 11 months 15 days left
296 Iter. #2: Thu 2024-02-29 00:00:00 UTC
297 From now: 4 years 11 months left
298 Iter. #3: Tue 2028-02-29 00:00:00 UTC
299 From now: 8 years 11 months left
300 Iter. #4: Sun 2032-02-29 00:00:00 UTC
301 From now: 12 years 11 months left
302 Iter. #5: Fri 2036-02-29 00:00:00 UTC
303 From now: 16 years 11 months left
304 systemd-analyze timestamp TIMESTAMP...
306 This command parses a timestamp (i.e. a single point in time) and
307 outputs the normalized form and the difference between this timestamp
308 and now. The timestamp should adhere to the syntax documented in
309 systemd.time(7), section "PARSING TIMESTAMPS".
310 Example 12. Show parsing of timestamps
312 $ systemd-analyze timestamp yesterday now tomorrow
314 Original form: yesterday
315 Normalized form: Mon 2019-05-20 00:00:00 CEST
316 (in UTC): Sun 2019-05-19 22:00:00 UTC
317 UNIX seconds: @15583032000
318 From now: 1 day 9h ago
319 Original form: now
321 Normalized form: Tue 2019-05-21 09:48:39 CEST
322 (in UTC): Tue 2019-05-21 07:48:39 UTC
323 UNIX seconds: @1558424919.659757
324 From now: 43us ago
325 Original form: tomorrow
327 Normalized form: Wed 2019-05-22 00:00:00 CEST
328 (in UTC): Tue 2019-05-21 22:00:00 UTC
329 UNIX seconds: @15584760000
330 From now: 14h left
331 systemd-analyze timespan EXPRESSION...
333 This command parses a time span (i.e. a difference between two
334 timestamps) and outputs the normalized form and the equivalent value in
335 microseconds. The time span should adhere to the syntax documented in
336 systemd.time(7), section "PARSING TIME SPANS". Values without units are
337 parsed as seconds.
338 Example 13. Show parsing of timespans
340 $ systemd-analyze timespan 1s 300s '1year 0.000001s'
342 Original: 1s
343 μs: 1000000
344 Human: 1s
345 Original: 300s
347 μs: 300000000
348 Human: 5min
349 Original: 1year 0.000001s
351 μs: 31557600000001
352 Human: 1y 1us
353 systemd-analyze cat-config NAME|PATH...
355 This command is similar to systemctl cat, but operates on config files.
356 It will copy the contents of a config file and any drop-ins to standard
357 output, using the usual systemd set of directories and rules for
358 precedence. Each argument must be either an absolute path including the
359 prefix (such as /etc/systemd/logind.conf or
360 /usr/lib/systemd/logind.conf), or a name relative to the prefix (such
361 as systemd/logind.conf).
362 Example 14. Showing logind configuration
364 $ systemd-analyze cat-config systemd/logind.conf
366 # /etc/systemd/logind.conf
367 ...
368 [Login]
369 NAutoVTs=8
370 ...
371 # /usr/lib/systemd/logind.conf.d/20-test.conf
373 ... some override from another package
374 # /etc/systemd/logind.conf.d/50-override.conf
376 ... some administrator override
377 systemd-analyze verify FILE...
380 This command will load unit files and print warnings if any errors are
381 detected. Files specified on the command line will be loaded, but also
382 any other units referenced by them. The full unit search path is formed
383 by combining the directories for all command line arguments, and the
384 usual unit load paths (variable $SYSTEMD_UNIT_PATH is supported, and
385 may be used to replace or augment the compiled in set of unit load
386 paths; see systemd.unit(5)). All units files present in the directories
387 containing the command line arguments will be used in preference to the
388 other paths.
389 The following errors are currently detected:
391 · unknown sections and directives,
393 · missing dependencies which are required to start the given unit,
395 · man pages listed in Documentation= which are not found in the
397 system,
398 · commands listed in ExecStart= and similar which are not found in
400 the system or not executable.
401 Example 15. Misspelt directives
403 $ cat ./user.slice
405 [Unit]
406 WhatIsThis=11
407 Documentation=man:nosuchfile(1)
408 Requires=different.service
409 [Service]
411 Description=x
412 $ systemd-analyze verify ./user.slice
414 [./user.slice:9] Unknown lvalue 'WhatIsThis' in section 'Unit'
415 [./user.slice:13] Unknown section 'Service'. Ignoring.
416 Error: org.freedesktop.systemd1.LoadFailed:
417 Unit different.service failed to load:
418 No such file or directory.
419 Failed to create user.slice/start: Invalid argument
420 user.slice: man nosuchfile(1) command failed with code 16
421 Example 16. Missing service units
424 $ tail ./a.socket ./b.socket
426 ==> ./a.socket <==
427 [Socket]
428 ListenStream=100
429 ==> ./b.socket <==
431 [Socket]
432 ListenStream=100
433 Accept=yes
434 $ systemd-analyze verify ./a.socket ./b.socket
436 Service a.service not loaded, a.socket cannot be started.
437 Service b@0.service not loaded, b.socket cannot be started.
438 systemd-analyze security [UNIT...]
441 This command analyzes the security and sandboxing settings of one or
442 more specified service units. If at least one unit name is specified
443 the security settings of the specified service units are inspected and
444 a detailed analysis is shown. If no unit name is specified, all
445 currently loaded, long-running service units are inspected and a terse
446 table with results shown. The command checks for various
447 security-related service settings, assigning each a numeric "exposure
448 level" value, depending on how important a setting is. It then
449 calculates an overall exposure level for the whole unit, which is an
450 estimation in the range 0.0...10.0 indicating how exposed a service is
451 security-wise. High exposure levels indicate very little applied
452 sandboxing. Low exposure levels indicate tight sandboxing and strongest
453 security restrictions. Note that this only analyzes the per-service
454 security features systemd itself implements. This means that any
455 additional security mechanisms applied by the service code itself are
456 not accounted for. The exposure level determined this way should not be
457 misunderstood: a high exposure level neither means that there is no
458 effective sandboxing applied by the service code itself, nor that the
459 service is actually vulnerable to remote or local attacks. High
460 exposure levels do indicate however that most likely the service might
461 benefit from additional settings applied to them.
462 Please note that many of the security and sandboxing settings
464 individually can be circumvented — unless combined with others. For
465 example, if a service retains the privilege to establish or undo mount
466 points many of the sandboxing options can be undone by the service code
467 itself. Due to that is essential that each service uses the most
468 comprehensive and strict sandboxing and security settings possible. The
469 tool will take into account some of these combinations and
470 relationships between the settings, but not all. Also note that the
471 security and sandboxing settings analyzed here only apply to the
472 operations executed by the service code itself. If a service has access
473 to an IPC system (such as D-Bus) it might request operations from other
474 services that are not subject to the same restrictions. Any
475 comprehensive security and sandboxing analysis is hence incomplete if
476 the IPC access policy is not validated too.
477 Example 17. Analyze systemd-logind.service
479 $ systemd-analyze security --no-pager systemd-logind.service
481 NAME DESCRIPTION EXPOSURE
482 ✗ PrivateNetwork= Service has access to the host's network 0.5
483 ✗ User=/DynamicUser= Service runs as root user 0.4
484 ✗ DeviceAllow= Service has no device ACL 0.2
485 ✓ IPAddressDeny= Service blocks all IP address ranges
486 ...
487 → Overall exposure level for systemd-logind.service: 4.1 OK 🙂
488
490 The following options are understood:
491 --system
493 Operates on the system systemd instance. This is the implied
494 default.
495 --user
497 Operates on the user systemd instance.
498 --global
500 Operates on the system-wide configuration for user systemd
501 instance.
502 --order, --require
504 When used in conjunction with the dot command (see above), selects
505 which dependencies are shown in the dependency graph. If --order is
506 passed, only dependencies of type After= or Before= are shown. If
507 --require is passed, only dependencies of type Requires=,
508 Requisite=, Wants= and Conflicts= are shown. If neither is passed,
509 this shows dependencies of all these types.
510 --from-pattern=, --to-pattern=
512 When used in conjunction with the dot command (see above), this
513 selects which relationships are shown in the dependency graph. Both
514 options require a glob(7) pattern as an argument, which will be
515 matched against the left-hand and the right-hand, respectively,
516 nodes of a relationship.
517 Each of these can be used more than once, in which case the unit
519 name must match one of the values. When tests for both sides of the
520 relation are present, a relation must pass both tests to be shown.
521 When patterns are also specified as positional arguments, they must
522 match at least one side of the relation. In other words, patterns
523 specified with those two options will trim the list of edges
524 matched by the positional arguments, if any are given, and fully
525 determine the list of edges shown otherwise.
526 --fuzz=timespan
528 When used in conjunction with the critical-chain command (see
529 above), also show units, which finished timespan earlier, than the
530 latest unit in the same level. The unit of timespan is seconds
531 unless specified with a different unit, e.g. "50ms".
532 --man=no
534 Do not invoke man to verify the existence of man pages listed in
535 Documentation=.
536 --generators
538 Invoke unit generators, see systemd.generator(7). Some generators
539 require root privileges. Under a normal user, running with
540 generators enabled will generally result in some warnings.
541 --root=PATH
543 With cat-files, show config files underneath the specified root
544 path PATH.
545 --iterations=NUMBER
547 When used with the calendar command, show the specified number of
548 iterations the specified calendar expression will elapse next.
549 Defaults to 1.
550 -H, --host=
552 Execute the operation remotely. Specify a hostname, or a username
553 and hostname separated by "@", to connect to. The hostname may
554 optionally be suffixed by a port ssh is listening on, separated by
555 ":", and then a container name, separated by "/", which connects
556 directly to a specific container on the specified host. This will
557 use SSH to talk to the remote machine manager instance. Container
558 names may be enumerated with machinectl -H HOST. Put IPv6 addresses
559 in brackets.
560 -M, --machine=
562 Execute operation on a local container. Specify a container name to
563 connect to.
564 -h, --help
566 Print a short help text and exit.
567 --version
569 Print a short version string and exit.
570 --no-pager
572 Do not pipe output into a pager.
573
575 On success, 0 is returned, a non-zero failure code otherwise.
576
578 $SYSTEMD_PAGER
579 Pager to use when --no-pager is not given; overrides $PAGER. If
580 neither $SYSTEMD_PAGER nor $PAGER are set, a set of well-known
581 pager implementations are tried in turn, including less(1) and
582 more(1), until one is found. If no pager implementation is
583 discovered no pager is invoked. Setting this environment variable
584 to an empty string or the value "cat" is equivalent to passing
585 --no-pager.
586 $SYSTEMD_LESS
588 Override the options passed to less (by default "FRSXMK").
589 Users might want to change two options in particular:
591 K
593 This option instructs the pager to exit immediately when Ctrl+C
594 is pressed. To allow less to handle Ctrl+C itself to switch
595 back to the pager command prompt, unset this option.
596 If the value of $SYSTEMD_LESS does not include "K", and the
598 pager that is invoked is less, Ctrl+C will be ignored by the
599 executable, and needs to be handled by the pager.
600 X
602 This option instructs the pager to not send termcap
603 initialization and deinitialization strings to the terminal. It
604 is set by default to allow command output to remain visible in
605 the terminal even after the pager exits. Nevertheless, this
606 prevents some pager functionality from working, in particular
607 paged output cannot be scrolled with the mouse.
608 See less(1) for more discussion.
610 $SYSTEMD_LESSCHARSET
612 Override the charset passed to less (by default "utf-8", if the
613 invoking terminal is determined to be UTF-8 compatible).
614 $SYSTEMD_COLORS
616 The value must be a boolean. Controls whether colorized output
617 should be generated. This can be specified to override the decision
618 that systemd makes based on $TERM and what the console is connected
619 to.
620 $SYSTEMD_URLIFY
622 The value must be a boolean. Controls whether clickable links
623 should be generated in the output for terminal emulators supporting
624 this. This can be specified to override the decision that systemd
625 makes based on $TERM and other conditions.
626
628 systemd(1), systemctl(1)
629systemd 243 SYSTEMD-ANALYZE(1)