WebCheck does not treat every transient failure as a confirmed outage. Normal network failures can be retried during a bounded retry window, and the incident policy then evaluates consecutive results before opening or recovering an incident.
Transient failures are retried
For eligible HTTP reachability failures, WebCheck retries at the configured retry interval during the retry window. A safety-policy block is not repeatedly bypassed. The final normalized result is stored as a check run.
When an incident opens
The configured failure threshold controls how many consecutive failing results are required before a normal incident opens. This helps separate a single network wobble from a persistent problem. Thresholds are configured on the target where supported.
When an incident recovers
The recovery threshold controls how many consecutive passing or degraded results are required before an open incident recovers. The incident timeline retains the opening, failure evidence and recovery information.
What you should inspect
Open the incident and compare the first failure with preceding runs and recovery runs. Check the error code, response status, timing, redirect chain and any certificate or content evidence before changing the target configuration.