Cron job monitoring that notices when a job stops running

Backups, imports and queue workers fail quietly. When a job stops starting, nothing raises an error in front of anyone, and a check from outside has nothing to connect to.

A heartbeat monitor waits for the job instead. Saving one gives you a private URL, and the job calls it when it finishes. You set how often a check-in is expected and how late it may be, and once both have passed without one, the monitor goes down.

The job can also call a /start address, so a run that hangs is caught once it passes its maximum runtime, and a /fail address to report a failure straight away. A new monitor stays off your public page until its first check-in, so a job you are still wiring up never shows as an outage.