Skip to content

systemd-analyze

Analyzes the boot process and unit files: what made the boot slow, in which order services start, and whether a unit file or a time specification is correct.

systemd-analyze output

Syntax

systemd-analyze [OPTION]... [COMMAND] [ARGUMENT]...

Options

Command Description
(none), time Total boot time, split into kernel, initrd and userspace
blame Start time of every unit, slowest first
critical-chain [UNIT] The chain of units that the boot or a unit had to wait for
verify FILE Check a unit file for errors
calendar EXPRESSION Check a time specification for timers and show the next run
security [UNIT] Rate the sandboxing settings of services

Examples

Show how long the boot took:

systemd-analyze
Startup finished in 1.912s (kernel) + 2.604s (initrd) + 5.317s (userspace) = 9.834s
multi-user.target reached after 5.301s in userspace.

Show which units took the longest to start:

systemd-analyze blame | head -n 5
4.211s postgresql@17-main.service
1.028s networking.service
 498ms php8.4-fpm.service
 351ms apparmor.service
 312ms nginx.service

Show what a service had to wait for:

systemd-analyze critical-chain nginx.service
The time when unit became active or started is printed after the "@" character.
The time the unit took to start is printed after the "+" character.

nginx.service +312ms
└─network-online.target @3.902s
  └─networking.service @2.871s +1.028s
    └─apparmor.service @2.512s +351ms
      └─local-fs.target @2.498s

@ is the point in time after boot, + the time the unit itself needed.

Check a time specification before you put it into a timer:

systemd-analyze calendar "Mon..Fri 03:00"
  Original form: Mon..Fri 03:00
Normalized form: Mon..Fri *-*-* 03:00:00
    Next elapse: Tue 2026-10-27 03:00:00 CET
       (in UTC): Tue 2026-10-27 02:00:00 UTC
       From now: 16h left

Check your own unit file:

systemd-analyze verify /etc/systemd/system/backup.service
backup.service: Command /usr/local/bin/backup.sh is not executable: No such file or directory

The script doesn't exist yet, or the path has a typo.

When do I need it?

  • When the server takes long to boot and you want to know which service is to blame
  • When a service starts too early, for example before the network is ready
  • Before enabling your own unit or timer, to catch errors in advance

blame isn't everything

Services start in parallel. A unit with a long time in blame doesn't necessarily delay the boot. critical-chain shows what the boot really waited for.

Test timers with calendar

systemd-analyze calendar also accepts --iterations=5 and then shows the next five runs. This lets you check complicated expressions like *-*-1,15 04:30.

systemctl · journalctl · uptime · systemd-run