Heartbeat monitors for cron jobs and backups

A heartbeat is a dead man's switch for anything that should run on a schedule: backups, queue workers, nightly imports.

Set it up

1. Add a monitor of type Heartbeat, choose the expected interval (e.g. every hour) and a grace period (e.g. 5 minutes).
2. Save. The monitor page shows a unique ping URL.
3. Make the job call that URL when it finishes – for cron, append && curl -fsS https://…/hb/TOKEN to the line.

What we do

We evaluate every minute. Once the interval plus grace has passed without a ping, the monitor is suspected down; three missed evaluations in a row confirm it and alert. Two fresh pings recover it. Heartbeats are evaluated by the app, never by a node, because only we know about the pings.