What Is a Dead Man’s Switch in Monitoring?

A dead man’s switch is a mechanism that triggers when expected activity stops. The name comes from trains and heavy machinery: a control the operator must hold; if they let go (or collapse), the machine stops safely. In software monitoring, the idea is the same — you expect a regular check-in, and absence is the alarm.

How it works

1. A job sends a small signal (a "ping" or "heartbeat") each time it runs successfully.
2. A monitor knows the expected interval and a grace period.
3. If a ping does not arrive in time, the monitor raises an alert.

Crucially, the alert is driven by what did not happen. That is what makes it reliable for scheduled work.

Why it beats watching the job directly

If you only watch the job for errors, you miss every case where the job never starts — a broken schedule, a dead host, a deploy that disabled it. A monitoring agent on the same server dies along with the server. A dead man’s switch lives elsewhere, so it survives total failure of the thing it watches.

When to use it

• Cron jobs and scheduled tasks (backups, billing, reports).
• Periodic batch pipelines and ETL jobs.
• Recurring queue workers and IoT device check-ins.
• Any process where "it silently stopped" is the failure you fear most.

A dead man’s switch you can set up in a minute

CronGuard is a hosted dead man’s switch for your jobs. Add a ping URL, choose the expected schedule, and get instant multi-channel alerts the moment a check-in is missed. No agent, free to start.

0 3 * * * /usr/local/bin/backup.sh && curl -fsS https://api.cronguard.dev/v1/ping/<your-uuid>