Check, sweep, open, feed.
A monitor is a URL and a check type. Everything downstream is automatic.
Three check types
HTTP(S) with status and body assertions, raw TCP connects, and ICMP ping. One monitor per endpoint you care about.
Every 10 seconds
The checker sweeps all monitors on a 10-second cycle. A degraded window shows up in under a minute, not after your users tweet about it.
Incidents run themselves
Consecutive failures open an incident; consecutive passes close it. No flapping alerts from a single dropped packet.
Critical, by definition
Uptime failures enter Incident Intelligence as critical — and merge with error and latency evidence pointing at the same cause.
Every degraded window, in context.
Each monitor carries a colour-coded 90-day availability timeline. Green is fine, amber is degraded, red is down — and every tick links to the checks behind it.
- Per-monitor 90-day history, colour-coded
- Response time and last-check freshness alongside
Opened by failures. Closed by passes.
Three consecutive failures open the incident; two consecutive passes close it. The thresholds keep a single flaky check from paging anyone — and the full check log stays on the incident.