I am required to set a postgres with geo-redundancy to a remote site.
I already have a local redundancy setup, using 1 standby for the Primary, and repmgr to ensure automatic failover in case the Primary crashes.
My thoughts were to add another standby on the remote site, so in case both the Primary and the standby crash on the local site, repmgr will switch over the primary to the standby on the remote site.
However seems like in case I manually crash both the primary and one of the standby's, then the second standby crashes and no swithcover is done.
Do I misunderstand something?
If Primary + one of the standby's crash - the second standby is of no use?