This is a case in which replica set failover doesn't work well - this would eliminate a single point of network failure.
- is duplicated by
-
SERVER-4377 handle most network partitions between mongos and primaries in replica sets
- Closed