From time to time, on all secondaries (even on not used by application),
cpu load peaks up to 100%, all used by mongo, the connections are hung.
Peak can last up to 30 minutes.
No obvious reason. I Thought, that it could be bad indexing, strange
queries, but it’s happening also on idling nodes…
Please note that there is no real idling unless no data is being written to
the replica set primary and all secondaries have no replication lag to
catch up on. You should also look for index build in the logs during this
The performance of any database system depends on several factors and might
not always have a proportional ratio to the number of documents. Size of
each document, types of queries running, hardware and software
configurations, other processes running on the machine, bad indexes, and
deployment architecture may all have a significant impact.
I hope this information helps you drill down to the exact factors that are
in effect during this time period as the graphs that you have put here are
not enough to justify the reasons behind bad performance of your
You received this message because you are subscribed to the Google Groups "mongodb-user"