One collector (of two) is having a high harvest time (great than 13 seconds). This causes the EM to "drop out". I have looked through and implemented many changes as per CA support and this board. The harvest duration still spikes. GC also spikes. One odd thing I noticed was the data points retrieved per interval for the one collector is over 1 million. It basically shows as a giant spike when the harvest duration spikes. The other collector is almost nil. What are the data points per interval and is this metric a clue as to what is going wrong?
The calculator harvest time also spikes (at 10,000ms) during the EM drop out.
This probably requires a deeper investigation and support is the right route but just quickly could you answer the following
- how many agents and metrics are on each collectors
- how many calc are on each collector
- what is the CPU and Heap Utilization on that collector
- In terms of hardware is it similar to the other