The connection shouldn’t drop once you’re connected, but if you don’t have the most recent ip list and they’ve turned off listening on that server you may be getting timeouts by the dns resolving to an old IP. I made sure to make all of that white noise on this iteration though, with the consolidation of the aws endpoints I was able to clean up my connection code a lot so there wasn’t a bunch of branching on ‘if x, y, z scenarios are true, do a, b, c’ - and instead it can focus solely on connect to a cluster server, and transparently migrate between available back ends based on the returned aws servers resolved.