Best way to restart LXD - final resort without having to reboot?


Recently been having frequent lockups where containers might go into ERROR state when just doing medial tasks, say lxc restart, not sure if they are related to dreaded namepsace kernel bugs though which require full reboot.

However, is there any last resort way of restarting some services that I could try before having to reboot the host?


Just tried to restart a container again and think it might be the namespace bug, not sure why I keep hitting it ?

commands I run in the lead up to this:

  882  lxc storage volume attach zfs3 plexstore3 emby1 plexstore3 /plexstore3
  883  #lxc config get plex1 volatile.idmap.base
  884  lxc config get plex1 volatile.idmap.base
  885  lxc config get emby1 volatile.idmap.base
  886  lxc config set emby1 security.idmap.base 1065536
  887  lxc config get emby1 volatile.idmap.base
  888  lxc config set plex1 security.idmap.base 1065536
  889  lxc config get plex1 volatile.idmap.base
  890  lxc restart plex
  891  lxc restart plex1
  892  lxc restart plex --force
  893  lxc restart plex1 --force
  894  lxc list
  895  lxc info --show-log plex1

Process stuck in D state forever is usually the sign of a kernel issue and you’re unlikely to be able to make it recover from that without a reboot.

You may have some more details in dmesg (kernel oops/crash/bug info)