We have a number of Datadog monitors which work wonderfully for identifying key issues in the system. We also have them integrated to PagerDuty, for alerting our teams and organizing responses.
This all works great, but the problem we're running into is the monitors are all set up with rules similar to "If x logs appear over y duration, alert", which kicks off a PagerDuty alert. However, after y duration (which can be very short), both the monitor and the PagerDuty alert are resolved, even though there may not have been enough time to respond.
How can I configure a monitor which will not automatically resolve, and requires manual intervention to move it back to the 'OK' state?