Updating to LXD 3.20 (source compiled) breaks CTs with dir backend, getting permission errors.
Downgrading breaks internal DB schema. Yay!
A DB backup is done on upgrades, so you can always revert that too.
Also, what errors are you getting?
Split this out as its own topic.
Sorry, broken from my end, not an LXD issue
3.20 seems to be running good. It will operate properly with two machines and will automatically swith masters in a 4 machine cluster when one goes down. And it will run find anywhere between 2 to 4 machines. It gets stuck with 1 machine.
I had to delete and reinstall from 3.19, upgrade left it half and half and would not work. It also still has issue coming up after reboot and needs lxd restart after units are up. I am thinking my server are trying to start lxd before network cards are full functional and they simply just get stuck and don’t try again unless I poke them. But the response in general is better.