Further documentation of postgres hosts in failover.md
Noted in 2018-06-21 failover
Currently we have a flat list of hosts for staging only in failover.md. At a minimum, we should add a flat list of hosts for production as well.
Since our replication is quite complicated and is multi-site to boot, I wonder if it might also be worth putting together mermaid diagrams covering the expected before and after state for both staging and production? I found it very useful to be able to understand why the web browser was showing one set of data while the rails console was showing another, for instance.
/cc @ibaum @jarv is this something one of you could take on? I'd also happily take static PNGs or google draws instead of mermaid diagrams :D