Hi,
I am monitoring a MongoDB 3.4 cluster running on KVM consisting of two shards with three nodes each plus the config server replica set plus two mongos processes. The cluster is managed by OpsManager and I am surprised how much impact the PMM monitoring seems to have. Attached are the OpsManager charts for the Operation Execution Times of a primary and a secondary node. As one can see the times did increase dramatically after the PMM clients (add mongodb) have been enabled. The PMM clients are on running on all of the nodes. Can anybody tell me if what the typical performance costs are? This seems much too high so I assume I must be doing something wrong or there is another problem of which I am (currently) not aware of…
Regards,
Bernd
