I think you are talking about an upgrade install. Those have a long history of breaking things. You would have to be crazy to attempt one of those on a critical production system.

What you would do for anything important is build a new separate system and then migrate to that once it is working. You can then migrate back if you discover issues too.

Yes. These sorts of upgrades were done on my home network, not an actual work-related system.

Thats also an argument for not using Ubuntu.

When I eventually rebuild things I’ll be going with Debian.

Unfortunately Debian wouldn't be exempt from "network interfaces renamed, lost default route, systemd timeouts taking 5 minutes, etc." because systemd is systemd (and Debian systemd maintainers just go along with upstream and do nothing to mitigate it)

I think interface names changed like at least 2 times in the last 4 releases.

Maybe? I run Debian on several VMs and my experience has proven otherwise. Major upgrades go a lot smoother than on Ubuntu.

Smoother, sure. I'm talking about interface name changes. it's nice you haven't encountered these releases but they sure happened in Debian, too.