Multi-region outage verification

4 min read

Single-region monitoring wakes you up at three in the morning for a false alarm because our ISP had a route flap. Multi-region solves this: an outage is confirmed only when reported by the primary AND at least one secondary worker in another region.

How it works

1. The primary check (from our main server) reports DOWN.
2. Before escalation, the system calls secondary workers (3-5 locations).
3. A worker performs a local test and returns UP/DOWN in 1-5 s.
4. If >= N workers confirm DOWN, the monitor is marked DOWN and alerts go out.
5. If the primary said DOWN but the workers say UP - it was a flake, no alert.

Current deployment status

Currently we have deployed one secondary worker - eu1 (EU region). The architecture is multi-region ready - adding more regions is prepared on the backend side and only requires deploying another worker node. We plan to expand in the coming months.

How it appears in the UI

In the monitor detail, for every DOWN event you see: