So personally I run Jelly and PLEX based on helper scripts LXC with pass trough. The rest of the arrs + qbit in a single Debian LXC (docker + portainer) as so lightweight. Thus only using 3 IPs (arrs + QB are just ports) YMMV
I’ve tried so many times to share the gpu with a lxc
No luck
I only snapshot my VMs anyway, the LXCs are so tiny that stop backups only cause an outage of a few seconds per container
There is a way to allow LXC snapshots. You don’t use mp in the LXC.conf, there’s another way to mount the NFS share. Let me know if you need the info, I can get it when I’m back in front of my computer. I was running and continuing with troubleshooting my *arr stack running on an unprivileged Alpine LXC with Docker. I just decided to migrate everything to a Debian VM with Docker. The unprivileged container didn’t allow for the container to write to the *arr /config directory due to folder ownership which I couldn’t change. I don’t feel comfortable running a privileged container, hence the change.