I have Oracle Enterprise Manager 12c installed managing a 2-node-RAC with multiple cluster databases. I recently added the recovery catalog, OEM repository and the listener on the corresponding (third) host as targets to the OEM.
In the Incident Manager I noticed the message Collection throttle removed with Severity None on the Listener target. This happens every 5 to 30 minutes.
When looking at the history of the message I can see that two alternating messages have alerting since I added the new targets.
- Collection throttle removed
- Resource manager has detected that the target's collections have exceeded
the expected collection time by _n_ milliseconds.
Subsequent collection of this target will be slowed down until
the collection evaluation is executing within its expected norm.
The second message has a severity Warning. The first message has no severity and happens everytime exactly 1 minute after the warning occurred.
While researching I came across a source saying that the collection is throttled whenever there is a high CPU usage. However, I cannot make anything of this because
- The average CPU usage is around 14% and the maximum rarely exeeds 30%.
- There are two other targets on the host (the repository catalog and the repository). If the cause was the CPU usage, wouldn't the other targets' collection also be throttled?