Sorry for the paleo post but curious if there was ever a resolution for
this issue regarding BDR member join/eject:
https://goo.gl/N7FoUB
I'm currently seeing very similar behavior for the latest version of BDR
where cluster nodes are very confused about the state of their
connections/replication slots when a new node joins with a completely
different local_node_name but same IP address as previous cluster member.
Thx