Edgedb 5 container generates too many connection errors and was suspended by cloud provider.

[BKG01]

• EdgeDB Version: edgedb query 'select sys::get_version_as_str()'
  "4.7+5f83595"
• EdgeDB CLI Version: EdgeDB CLI 5.1.0+7c5764f
• OS Version: Linux - (node bullseye slim0

Steps to Reproduce:

1. Follow the edgedb official instructions for a v4 deployment to azure (including setting up a flexible postgres)
2. Then tear down the v4 container and deploy a v5 container instance (connected to the existing Postgres backend)

We had a recent issue where our cloud provider(Azure) suspended (put in a forced waiting state) our edgedb container as it was behaving badly (too many reconnection attempt). The edgedb container image specified wasn't specific enough so when the container restarted it pulled down the latest version of edgedb 5(.2) whilst we hadn't yet migrated from v4(.7) which we think was the root of the issue with the edgedb container. As it was suspended we couldn't access any logs etc... so it was hard to diagnose what the issue was. It would be really helpful if a v5 edgedb container when connected to a postgres instance that hasn't migrated yet, doesn't thrash the db with connection attempts.

Note there are no logs available to post from the v5 container as is was suspended and the logs were inaccessible.

[CarsonF]

This could be similar to what I'm seeing in AWS, where RDS closes idle connections and then edgedb re-establishes them. Every 4 minutes. Though I don't think it's hurting anything other than log noise for us.