|
SERVER-32270 would have avoided this issue, since heartbeats and data replication would have been sharing the same connection pool and thus if heartbeats could get a connection, data replication would also be able to get a connection.
There's a broader conversation to be had here about what it means for a node to report itself as "alive". A node that can heartbeat but cannot replicate data, for any reason, should probably not be considered "alive". Figuring out a story for tying liveness monitoring in with the ability to replicate data would be very valuable, but is complex and would be a significant change to the current architecture. It's something I'd love to do, but I don't expect us to get to it in our current 3.7 development cycle.
For now, I'd like to close this ticket as a duplicate of SERVER-32270, which would address the specific networking issues brought up in this ticket, and I will file a separate ticket for a more comprehensive story around liveness monitoring.
|