Enable postgres replication between Coordinating Nodes
The goal of this story is to enable replication of postgres databases between Coordinating Nodes.
The replication process should:
- be robust across a wide area network with potentially high latency.
- recover consistency after being disconnected deliberately or otherwise
- Assume a single master, with at least two, but potentially several slaves
- Support ease of switching the master to one of the slaves (e.g. switching from cn-ucsb-1 to cn-orc-1)