SystemD

Services has a new name... happy! They are now units

Look also:

• Official docs
• SystemD for Upstart Users
• RedHat docs

Paths

Central path system-wide

/lib/systemd/system     # debian-based
/usr/lib/systemd/system/    # redhat-based

Put general custom units in

/etc/systemd/system/

Override a systemd profile

To override some attributes inside example.service, create

/etc/systemd/system/example.service.d/mycustom.conf

and write there you customizations

System Logs

Show current logs

journalctl -n         # show onetime
journalctl -f         # tail logs

Show today logs from two units:

journalctl -u nginx.service -u php-fpm.service --since today

Show kernel logs

journalctl -k                # last boot
journalctl -k -b -5          # five boots ago

Show by field (man systemd.journal-fields)

journalctl _UID=33 --since today

Show only error messages, from boot

journalctl -p err -b

Other interesting options

--no-pager        # don't use more
--no-full         # truncate at last column
-a                # show all, even non printable chars

Simple service definition

[Unit]
Description=ROT13 demo service
After=network.target
After=mysqld.service
StartLimitIntervalSec=0

[Service]
Type=simple
Restart=on-failure
RestartSec=2        # restart after 2 seconds (default 100ms)
User=centos
ExecStart=/usr/bin/env php /path/to/server.php

[Install]
WantedBy=multi-user.target

See also Unit Manual

Service Auto-Restart

Example 1: try a restart per second and stop after 5 attempts in 10 seconds

[Unit]
StartLimitBurst=5       # default
StartLimitIntervalSec=10    # default

[Service]
Restart=always
RestartSec=1        # default: 100ms

Example 2: try a restart every 30 seconds and stop after 5 attempts in 5 minutes reboot after 5 failed attempts to restart the service

[Unit]
StartLimitBurst=5       # default
StartLimitIntervalSec=5min  # default
StartLimitAction=reboot

[Service]
Restart=always
RestartSec=30       # default: 100ms

Note that systemctl reset-failed will cause the restart rate counter for a service to be flushed, which is useful if the administrator wants to manually start a unit and the start limit interferes with that.

Manage process limits

Check current ulimits for an hypotethical MongoDB

root@xenial:~# systemctl --no-pager show mongodb.service | egrep 'NOFILE|NPROC'

Change values

root@xenial:~# cat /etc/systemd/system/mongodb.service
[Service]
LimitNOFILE=64000
LimitNPROC=infinity

root@xenial:~# systemctl daemon-reload