Heartbeat monitoring
Heartbeat monitoring expects a scheduled job to check in, and alerts you when the expected ping doesn't arrive.
Also called a dead man's switch, heartbeat monitoring inverts the usual model. Instead of a monitor polling your service, your service pings the monitor on a schedule. As long as the expected check-ins keep arriving on time, all is well; when one is late or missing, the monitor raises the alarm. It is the natural fit for things that run rather than things that serve.
Why it matters: many critical jobs have no endpoint to poll — backups, cron jobs, queue workers, data pipelines, scheduled exports. Traditional uptime checks can't see them. When one of these fails silently, you often discover it only when the missing output is needed (an empty backup, a stale report). Heartbeats turn that silent failure into an immediate, explicit alert.
A common misconception is that a heartbeat only catches a job that crashes. A well-configured heartbeat with a grace period also catches a job that hangs, runs far too long, or never starts because a scheduler is broken — anything that breaks the expected rhythm of check-ins, not just an outright crash.
In practice you give each job a unique heartbeat URL and an expected interval plus grace window. The job calls that URL on success; if the monitor doesn't hear from it in time, it alerts through your normal channels. This is the standard way to monitor work that produces no continuously pollable surface.
Frequently asked questions
- What is heartbeat monitoring used for?
- It monitors scheduled or background work that has no endpoint to poll — cron jobs, backups, queue workers, and data pipelines — by expecting each to check in on schedule and alerting when one goes silent.
- How does a heartbeat differ from an uptime check?
- An uptime check actively polls a service from the outside. A heartbeat is passive and inverted: your job pushes a ping to the monitor, and the absence of that ping is what triggers the alert.
- What is a grace period?
- A grace period is extra time allowed beyond the expected interval before a missed heartbeat counts as a failure, so a job that occasionally runs a little long doesn't trigger a false alarm.
Start monitoring in under a minute
Free forever, no credit card. Upgrade when you need finer intervals.