Day 15 - Scheduling with cron and systemd timers

Linux offers two standard ways to run jobs on a schedule: classic cron and modern systemd timers. This lesson explains how to create reliable schedules, control environment and logs, avoid overlapping runs, and verify results.

Prerequisites

• Day 14 on shell scripting
• A terminal with sudo privileges

Cron basics

Each user can have a personal crontab. The cron daemon runs jobs in minimal environments.

# edit user crontab
crontab -e
 
# list current entries
crontab -l
 
# system crontab and directories
sudo ls -l /etc/crontab /etc/cron.*

Crontab entry format:

# ┌ minute (0-59)
# │  ┌ hour (0-23)
# │  │   ┌ day of month (1-31)
# │  │   │  ┌ month (1-12 or names)
# │  │   │  │  ┌ day of week (0-7 or names, 0 and 7 are Sunday)
# │  │   │  │  │
# *  *   *  *  *  command

Special strings:

• @reboot, @yearly, @annually, @monthly, @weekly, @daily, @hourly

Examples:

# run a script every day at 02:30
30 2 * * * /home/user/bin/backup.sh >>/home/user/logs/backup.log 2>&1
 
# run every 15 minutes
*/15 * * * * /home/user/bin/healthcheck.sh >/dev/null 2>&1
 
# run at boot
@reboot /home/user/bin/warmup.sh

PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
SHELL=/bin/bash
MAILTO=""

Cron logging and mail

• Anything written to stdout or stderr is mailed to the user unless MAILTO is empty or output is redirected
• On many systems, cron logs appear in /var/log/syslog or /var/log/cron

Redirect to a file and keep both streams:

0 3 * * * /home/user/bin/report.sh >>/home/user/logs/report.log 2>&1

anacron for machines that sleep

Laptops may miss 02:00 if powered off. anacron runs daily, weekly, and monthly jobs at the next opportunity.

sudo apt install -y anacron 2>/dev/null || sudo dnf install -y cronie-anacron 2>/dev/null

System jobs for anacron live under /etc/anacrontab and /etc/cron.*.

Prevent overlapping runs with flock

Use a lock to ensure only one instance runs at a time.

# crontab line, runs every 5 minutes but skips if a previous run is active
*/5 * * * * /usr/bin/flock -n /run/myjob.lock /home/user/bin/myjob.sh >>/home/user/logs/myjob.log 2>&1

-n makes flock fail fast if the lock exists. Use -w 30 to wait up to 30 seconds instead.

Systemd timers overview

Timers pair a .timer schedule with a .service that does the work. Benefits include integration with journalctl, boot catch up with Persistent=true, jitter with RandomizedDelaySec, and easy inspection with systemctl list-timers.

Create files under /etc/systemd/system/ for system wide timers, or under ~/.config/systemd/user/ for per user timers.

Example: daily backup at 02:30 with jitter and boot catch up

Create the service unit:

# /etc/systemd/system/backup.service
[Unit]
Description=Daily backup
 
[Service]
Type=oneshot
ExecStart=/usr/local/bin/backup.sh
# pass environment or load from a file
Environment=BACKUP_TARGET=/srv/backups
EnvironmentFile=-/etc/default/backup

Create the timer unit:

# /etc/systemd/system/backup.timer
[Unit]
Description=Run backup daily at 02:30
 
[Timer]
OnCalendar=*-*-* 02:30:00
Persistent=true
RandomizedDelaySec=15m
AccuracySec=1m
 
[Install]
WantedBy=timers.target

Enable and start:

sudo systemctl daemon-reload
sudo systemctl enable --now backup.timer
systemctl list-timers | grep backup

Logs and last run:

systemctl status backup.service
journalctl -u backup.service -n 50 --no-pager

Calendar expressions

Examples for OnCalendar:

• daily, weekly, hourly
• Mon..Fri 09:00
• *-*-01 03:15:00 on the first day of each month
• 2025-10-05 22:00
• Sun *-*-* 01:30:00

Non overlapping runs with systemd

Use a lock inside the service or let systemd refuse a second start while one is running.

# /usr/local/bin/myjob.sh called by myjob.service
#!/usr/bin/env bash
set -euo pipefail
exec /usr/bin/flock -n /run/myjob.lock /usr/local/bin/myjob-real.sh

Systemd will queue another start if one is in progress. The flock pattern prevents concurrent work.

User timers without sudo

Per user timers live in ~/.config/systemd/user/ and run under the user session.

mkdir -p ~/.config/systemd/user
 
cat > ~/.config/systemd/user/notes.service <<'EOF'
[Unit]
Description=Append timestamp to notes
 
[Service]
Type=oneshot
ExecStart=/usr/bin/sh -c 'date +%F" "%T >> "$HOME/notes.log"'
EOF
 
cat > ~/.config/systemd/user/notes.timer <<'EOF'
[Unit]
Description=Every 15 minutes
 
[Timer]
OnCalendar=*:0/15
Persistent=true
 
[Install]
WantedBy=timers.target
EOF
 
systemctl --user daemon-reload
systemctl --user enable --now notes.timer
systemctl --user list-timers
journalctl --user -u notes.service -n 20 --no-pager

Time zones and time sources

Confirm system time and time zone.

timedatectl

If jobs appear to run at the wrong time, check NTP status and the selected zone. Systemd timers use the system time zone by default.

Practical lab

1. Create a user crontab that writes a timestamp to ~/cron-demo.log every 2 minutes. Add a safe PATH line at the top and test that logs capture output.

2. Add flock to a cron entry that runs every minute, then sleep for 120 seconds inside the script to confirm that a second run is skipped while the first is active.

3. Build a cleanup.service and cleanup.timer pair that removes files older than 14 days from ~/tmp every day at 01:15. Use RandomizedDelaySec=10m.

4. Switch to a per user timer by moving the units into ~/.config/systemd/user/ and enabling them with systemctl --user.

5. Verify a missed run after reboot by stopping the timer for a day, setting the schedule to a past time, enabling Persistent=true, and checking the journal.

Troubleshooting

• Cron job runs in the terminal but not in cron. The environment is different. Set PATH, use absolute paths, and refer to scripts with full paths.
• No cron mail. Install an MTA or set MAILTO="" and redirect output to a log file.
• crontab: installing new crontab shows but jobs never trigger. Confirm the cron service name on the distribution and start or enable it: sudo systemctl status cron or sudo systemctl status crond.
• Systemd timer did not fire. Inspect with systemctl list-timers, check systemctl status <name>.timer, and read logs with journalctl -u <name>.service.
• Job overlaps or double runs. Wrap the script with flock for cron and systemd services.
• Wrong time. Check timedatectl, verify NTP sync, and confirm the time zone.

Next steps

Day 16 covers service management with systemd. It explains units, status, enable and disable, logs, dependencies, targets, and safe patterns for writing custom services.