In an unpredictable maintenance window environment, as in, SQL Monitor's abilities in the window setting arena are inadequate, there are a reasonable number of times when SQL Agent is restarted with good reason.

As alerts are tied to pagerduty, it's a one size fits all Alert at present, which isn't helpful.

If a "delay emitting Alert for X time periods" could be added to this Alert, the short-duration bounces will go un-alerted, which really helps on call personnel to avoid unnecessary awakenings for "it is of no matter" Alerts.

In my case, there are 24 recorded incidents of this Alert, absolutely none of which required DBA intervention. Most of these ended in under 10 minutes. All ended under 30 minutes.

A 3-stage low, medium, high, with override email address allocation AT EACH LEVEL, would be ideal. For low, I can nudge just the DBA Team, for Medium, if 10 minutes, I can nudge again, but for > 30 minutes, PagerDuty.