The most dangerous failure
makes no noise.
A process dies, a cron stops firing, a queue consumer stalls. There are no error logs — just absence. Threshold alerts can't fire on nothing. Epok notices when a service that normally talks goes quiet.
Every alerting setup is built to catch things happening: errors, latency, saturation. Almost none catch things stopping. When a worker silently dies or a scheduled job misses its run, the error rate doesn't spike — it goes to zero, which looks exactly like a quiet, healthy night.
You find out hours later, when the downstream that depended on it finally breaks loudly — and now you're debugging the symptom three services away from the cause.
Absence is a signal
Epok learns each service’s normal volume and flags when one that logs steadily falls quiet — the failure mode that produces no error to alert on.
Seasonality, not naive thresholds
It knows a service’s quiet hours. A 3am lull on a daytime-only job isn’t a page; a daytime job going dark at noon is.
Scoped to the service
The alert names the stream that went silent and the baseline it deviated from — you start at the source, not the downstream wreckage.
No rules to write
You don’t author an absence check per service. Detection learns what steady looks like and watches for its disappearance automatically.
- ✓A page when a normally-active service stops logging — before the downstream notices.
- ✓The expected baseline it fell below, so the alert is justified, not jumpy.
- ✓Awareness of normal quiet periods, so weekends and off-hours don’t generate false pages.
- ✓The service named at the source of the silence, not the cascade it eventually triggers.
Dead crons, stalled consumers, and crashed workers stop being the incident you find out about from a customer.
Watch it catch this one. No signup.
The live demo runs detection on real-shape data — every alert cited back to the line behind it. Or send your own logs in the 14-day trial.