TCP keepalive options in MongoDB Diagnostics FAQ

The Diagnostics section indicates that the TCP keep alive setting should be the same for all components. While settings for the mongoD process should be consistent across all nodes, it is possible to have a different setting on mongoS and or client machines.

This is not clear from the Diagnostic section of the documentation as it stands.

If your MongoDB deployment experiences keepalive-related issues, you must alter the keep alive value on all machines hosting MongoDB processes. This includes all machines hosting mongos or mongod processes and all machines hosting client processes that connect to MongoDB.

This statement reads that the keep alive value must be identical on all mongoS, mongoD, and client machines.