cron and systemd timers: run something every night

You have a script that works when you run it by hand and you want it to run at 02:30 every night. Linux gives you two standard tools: cron, which is simple and everywhere, and systemd timers, which give you logging, dependencies and catch-up of missed runs for free. This guide sets up the same nightly job both ways.

Before you start

• The script runs with no terminal and a nearly empty environment. Use absolute paths inside it and start it with set -euo pipefail.
• Make it executable: chmod +x /usr/local/bin/nightly-report.sh.
• Test it once from a clean shell: env -i HOME="$HOME" PATH=/usr/bin:/bin /usr/local/bin/nightly-report.sh. If that fails, cron will fail the same way.

Option 1: cron

Each user has a crontab. Edit yours with crontab -e (it opens in your editor) and list it with crontab -l. A line has five time fields followed by the command.

# min  hour  day-of-month  month  day-of-week   command
  30   2     *             *      *             /usr/local/bin/nightly-report.sh >> /var/log/nightly-report.log 2>&1

• * means every value. */15 in the minute field means every 15 minutes. 1-5 in day-of-week means Monday to Friday. 0,30 lists values.
• Shortcuts: @daily (midnight), @hourly, @weekly, @reboot.
• Set [email protected] at the top to get any output by mail, or redirect output to a log file as above. A job that prints nothing and exits 0 is silent.
• Set PATH=/usr/local/bin:/usr/bin:/bin at the top of the crontab; the default PATH in cron is minimal, which is why command not found is the most common cron failure.
• The % character is special in crontab lines (it becomes a newline). Escape it as \%, for example in date +\%F.

System-wide jobs that must run as root go in /etc/cron.d/ as separate files with an extra user field before the command, or you simply drop an executable script into /etc/cron.daily/.

# /etc/cron.d/nightly-report
PATH=/usr/local/bin:/usr/bin:/bin
MAILTO=root
30 2 * * * root /usr/local/bin/nightly-report.sh >> /var/log/nightly-report.log 2>&1

Check that it ran: look at your log file, or at the cron daemon’s log with journalctl -u cron (Debian, Ubuntu) or journalctl -u crond (RHEL, Fedora), or grep CRON /var/log/syslog on older systems.

Option 2: a systemd timer

A timer is two small files: a .service that says what to run and a .timer that says when. Put them in /etc/systemd/system/.

# /etc/systemd/system/nightly-report.service
[Unit]
Description=Nightly report

[Service]
Type=oneshot
ExecStart=/usr/local/bin/nightly-report.sh
User=report
Nice=10

# /etc/systemd/system/nightly-report.timer
[Unit]
Description=Run nightly report at 02:30

[Timer]
OnCalendar=*-*-* 02:30:00
Persistent=true
RandomizedDelaySec=5m

[Install]
WantedBy=timers.target

sudo systemctl daemon-reload
sudo systemctl enable --now nightly-report.timer

systemctl list-timers --all                 # next and last run of every timer
sudo systemctl start nightly-report.service  # run the job right now, by hand
journalctl -u nightly-report.service -n 50   # its output, no redirection needed

• Persistent=true runs the job at boot if the machine was off when it was due. cron simply skips missed runs.
• RandomizedDelaySec spreads jobs on many machines so they do not all hit a server at the same second.
• Check a calendar expression before you rely on it: systemd-analyze calendar "Mon..Fri 02:30" prints the next run time.
• Shortcuts such as OnCalendar=daily, weekly and hourly work too.

Per-user timers without root go in ~/.config/systemd/user/ and are managed with systemctl --user. They only run while you are logged in unless you enable lingering with loginctl enable-linger $USER.

Which one to pick

• cron: a one-liner on a machine you do not own or that is not running systemd, or a quick personal job.
• systemd timer: anything on a server you maintain. You get journalctl logs, catch-up of missed runs, resource limits and systemctl status for free.

Whichever you choose, make the script itself safe to be started twice and loud when it fails. The cron-safe wrapper script in this encyclopedia adds locking, logging and failure reporting around any command.

Official documentation

• crontab(5) manual page
• systemd.timer
• systemd.time: calendar event syntax