Suppress alerts for a short time, then resume alerting if the check is still failing

Use case

We monitor a fleet of customer-facing hosts with one URL monitor per host. A new host takes a minute or two to start, so its first runs fail. We want to hold back alerts for the first minute after a check is created (or while we deploy), and still be alerted if the host is down once that period is over.

What we found

1. A maintenance window with silenceAlertsTags silences the alert, as expected.

2. If the check starts failing inside the window and is still failing when the window ends, no alert is ever sent. Checkly alerts on a state change, and that change happened inside the window. We saw exactly one ALERT_FAILURE recorded (silenced) and nothing in the following five minutes.

3. Alert reminders do not help: there is no delivered alert for them to repeat.

4. Support AI chat confirmed this is by design and that nothing notifies at the end of a window.

The result is that a window can hide a host that is broken from the start, and it stays hidden until the host recovers and fails again.

What we would like

Either:

• a "silence, then resume" behaviour: when a window ends and a check is still failing, send the failure alert (or let reminders start); or

• a grace period on a check or alert channel, for example "do not alert for the first N minutes after the check is created"; or

• a notification when a window ends and checks are still failing.

Our workaround

We now require two failed runs in a row (failedRunThreshold: 2). That gives new hosts about 90 seconds and still alerts on a continuing failure, but it delays alerts for real outages on established hosts by a minute.