Cron job monitoring: know when a job did not run

Why cron jobs fail quietly

• The server was rebooted, the crontab was lost in a deploy, or the cron daemon stopped.
• The script exits early with an error, and the output goes to a mailbox nobody reads.
• The job hangs, or takes so long that it overlaps the next run.
• A scheduled CI workflow got disabled (GitHub disables scheduled workflows in inactive repositories).

Checking a website from outside cannot catch any of these. You need the job to report in.

How heartbeat monitoring works

1. You create a heartbeat with a period (how often the job runs) and a grace time (how late is still fine).
2. You add one line to the end of the job that requests the heartbeat URL, only when the job succeeded.
3. If period plus grace passes with no ping, the heartbeat is marked late and you get one alert. The next ping closes it and sends one recovery alert.

Pick the grace time from how long the job usually takes plus some slack. A nightly backup that takes 20 minutes might use a 1 day period with 1 hour grace.

Heartbeat quick start

1. Create a free Pingmoss account (no email needed) and add a cron heartbeat. Copy its ping URL.
2. Append the ping to your job, using && so it only runs on success:

5 3 * * * /usr/local/bin/backup.sh && curl -fsS -m 10 --retry 3 https://pingmoss.pages.dev/hb/YOUR_TOKEN > /dev/null

3. Add an alert channel (webhook, Discord, Slack or Telegram) and send a test alert.

Examples for GitHub Actions, systemd, Python and Node are in the heartbeat docs. The Free plan includes 5 monitors, and heartbeats can use any period from 1 minute to 31 days. Pingmoss does not send email alerts yet and has no start or fail signals, so it covers the "did it run and succeed" question only. Comparing tools? See Pingmoss vs UptimeRobot, Healthchecks.io and Uptime Kuma.