Retries and pausing

A 2xx status is a success. Everything else is a failure: any other status, a DNS failure, a refused connection, a TLS failure, or a timeout. A redirect is recorded with its status and not followed.

The ladder

A failed run is retried after one minute, then five, then thirty. A fourth consecutive failure pauses the job. A success at any point clears the counter.

A server asking for a specific wait through Retry-After gets it, as long as the wait is half an hour or less. When the ordinary schedule comes round sooner than the retry would, the scheduled run carries the attempt instead of stacking a retry in front of it.

Pausing

A paused job keeps its schedule and its history, and resumes from the dashboard or from the link in the pause email. The pause reason names the likely cause, so a 301 reads as a moved URL rather than a bare number.

Runs the service could not attempt are recorded as unknown and never count toward a pause.