How Mage Watcher decides a site is down

Every monitor is checked on its own schedule from the monitoring regions you choose for it. A check succeeds when the site answers with 200 OK within the timeout and, if you set Expected content, the page contains that text.

A single failed check does not raise the alarm. Network problems near one region happen, and they are not your site being down. Instead:

1. A region sees a failure. Mage Watcher opens an incident in the Investigating state. Nobody is notified yet, and the public status page still shows the service as operational.
2. Other regions are asked to check. Mage Watcher immediately asks additional regions to check the same address.
3. The regions agree, or they do not. If at least two regions see the failure, the incident becomes Confirmed, the monitor shows as Down, and your alerts go out. If another region sees the site working, the incident is closed as a False alarm: no alert, and no effect on your uptime.
4. When no other region can check. If no other region is available to verify, Mage Watcher judges on the first region's own record instead. Two failures in a row from that region confirm the outage; a success from any region since the failure makes it a false alarm. If no further result arrives at all, it is closed as a false alarm because it could not be verified.
5. Recovery is verified too. When a check succeeds again during a confirmed outage, the incident moves to Monitoring and other regions are asked to confirm the site is back. Once they agree, the incident is Resolved and the "back up" alert goes out. If a recovery check fails, the incident goes back to Confirmed.

Each step is written to the incident's Timeline, so you can always see how a decision was reached.