Backups, billing runs, digests, syncs. When a scheduled job stops, nothing errors and nobody notices. Fajita listens for the jobs that stop reporting in.
heartbeat nightly-backup · expected every 24h→ last ping 26.2h agograce window 2h · exceededincident opened · 'nightly-backup went quiet'your script's side: curl -fsS https://hb.fajita.io/p/…
What it checks
Each job gets a private heartbeat URL and an expected schedule. The job pings when it finishes; Fajita watches the clock.
What failure looks like
Nothing. That is the problem. A dead cron job produces no error, no log line, no page. Just an absence that compounds daily.
What Fajita does
A missed ping past its grace window counts as a failure: verified, alerted, and recorded like any outage. A resumed ping closes the incident.
Why it matters
The nightly backup that stopped in March is discovered during the restore in July. Heartbeats turn that discovery into a same-day alert.
In practice
Fair questions
No. Your job makes an outbound request to its heartbeat URL. Fajita never connects into your infrastructure.
Every heartbeat has a grace window. A job that is occasionally slow will not page anyone; a job that is absent will.
Related