60% mongod-CPU Usage on IDLE

Using MongoDB 3.2.4 an ElasticSearch 2.3.1 I’m experiencing on IDLE a CPU load of ~60% from a mongod-process. The mongoDB-log is filled every 3 seconds with such an entry: 2016-05-24T01:32:54.296+0200 I COMMAND [conn598] command local.oplog.rs command: count { count: "oplog.rs", query: { ts: { $gte: Timestamp 1462970034000|1 } } } planSummary: COLLSCAN keyUpdates:0 writeConflicts:0 numYields:35407 reslen:62 locks:{ Global: { acquireCount: { r: 70816 } }, Database: { acquireCount: { r: 35408 } }, oplog: { acquireCount: { r: 35408 } } } protocol:op_query 1688ms

So I assume, that this is the mongo-connector, who is polling the oplog(-timestamp) from mongoDB. There are two questions now:

1. Why does this simple query needs so much CPU?
2. How to decrease the needed IDLE-ressources? First I’d think of decreasing the polling-rate for example to one minute. But how? I hope I didn’t miss the corresponding config-option on my research.

[behackett]

Actually, in your case we’re just querying on the ts field, which should be fast since we use the oplogReplay option. This requires more research.

[mumlax]

I updated my mongo-connector instance successfully. Until now I registered a decrease to an average CPU Usage of 10% by mongod on IDLE. Sometimes peaks up to 40%. Thats a significant improvement, I can confirm the issue-resolution! ✔️ Thanks a lot for the fast fix!