We have set up a blocking report to trigger every 1 min in DPA using the generic alerts
Total Blocking Wait Time
and one using
Single Resource Metric/blocked sessions
My question is as follows:
For these alerts, do they query the repository to see what blocking has occurred and already been captured by DPA for the past minute...or are they actively monitoring the live DB server and at that moment in time (the 1 min check) checking for blocking and reporting on it.
If you are using the canned resource for blocked sessions, then, yes, we query the repo for data already collected. This offloads any overhead on your monitored instance as the data has already been collected. Having said that, the default collection for that resource is in 1 minute intervals. Setting up the alert to also be for 1 minute might be a little tight?... Have you considered every 5 or 10 minutes to check if blocking had occurred?