@MacFan4000 You lucked out and there was a backup that predated the VM getting scrambled; I've restored it to the state it was in on June 20th. I can't make any promises about how postgres will or won't feel about the time traveling.

A few notes for future reference:

1. Don't do in-place OS upgrades of your VMs. It often leads to disaster (as seen today) and in any case results in hosts that are in an inconsistent/mysterious state and difficult to maintain. Among other things, a VM created with a Buster base image will always appear forever after in any report we run about Buster VMs, which means it's likely to get arbitrarily deleted when we deprecate the old OS version.

2. When you replace this VM with a new one, you'll want to move your data onto a cinder volume; that means the data will be preserved and recoverable regardless of what happens to the OS of the host.

3. Trove (or dbaas offering) has partial Postgres support, so you might also want to experiment with that and see if it's a good alternative.

Good luck!

-A