I have a client who uses Proxmox and its backup server. Last week, I upgraded the backup server from Debian 12 to 13. The backup server "sleeps" most of the day, so it also runs Docker for a Gitea runner. Everything seemed fine initially.
Then, my client messaged me yesterday because the runner had stopped working. When I logged in, I found that for some reason, the runner could no longer connect to the Docker socket, even though I was passing it the official way. I tried the same thing on a different Debian 13 server and got the same result. But, on a Debian 12 VM using the (old) Docker from the Debian repos, everything worked perfectly.
This incident just reinforces my point that for production servers, it’s crucial to use solutions that don't introduce breaking changes between releases. It seems to be an Apparmor issue (thanks @gyptazy for the head up!).
Because this component was non-critical and easily replaceable, I didn't pay much attention to testing it right after the server upgrade.
