The issue we have is that the cutover is super hard to do in a zero-downtime way because we almost always have active connections and you can’t have multiple pr

The issue we have is that the cutover is super hard to do in a zero-downtime way because we almost always have active connections and you can’t have multiple primaries. There’s tricks like read connections on the replicas but it’s still really hard to coordinate. Given that Postgres has built in synchronous replication, I feel like it should also have some support for multiple primaries (during a cutover window) to allow better HA support