The only item we can see in the events page for the volume is
Removing unknown replica
tcp://10.138.31.19:10694 in mode ERR from engine
The volume will happily operate with an additional replica indefinitely until we decrease the replica count so don't think the new replica itself is the issue