We had an instance go DOWN last night .. but OEM12c never triggered the critical alert. Looking into the metric values we see:
We got the Warning alert for MOUNTED but never got an alert for the database being down.
The metric policy is set to DOWN|UNKNOWN - should that have fired? Should I change the metric to Down|DOWN|UNKNOWN or is something wrong with the metric evaulation?
Yes you right
When using the Matches operator for a metric, the values defined in the Critical or Warning Thresholds are case sensitive, so the match must be exact.
Add the value "Down" to the Critical threshold of the Response:State metric so that the match set of values becomes "DOWN|UNKNOWN.*|Down" on the Metric page of the database target.
Once done it should work