In production after automatic refresh to 3.14, it’s not starting,
running lxd in debug shows:
EROR[06-19|15:35:49] Failed to start the daemon: open /var/snap/lxd/common/lxd/storage-pools/btrfspool/containers-snapshots: is a directory
After i removed the containers-snapshot dir i get then:
INFO[06-19|15:46:09] Applying patch: storage_api_rename_container_snapshots_dir_again
EROR[06-19|15:46:09] Failed to start the daemon: open /var/snap/lxd/common/lxd/storage-pools/btrfspool/containers-snapshots: is a directory
I have got the exact same problem with my ‘default’ storage-pool, also using btrfs-driver.
Making it a bit more confusing (for me at least) because the symlink usually found under ‘/var/snap/lxd/common/mntns/’ no longer points to a valid destination, as the lxd-daemon no longer runs.
The fix for this issue is now available in the stable channel, run snap refresh lxd to force the update.
The source of this seems to be:
- Systems with btrfs storage pools
- Snapshots on containers
- A previous failure during a storage update which failed to fully move all snapshots
So it’s somewhat unlikely to be hit but we’ve clearly seen 5-6 people so far hit this, the updated snap has a fix we included upstream this morning and which goes through all containers and snapshots and moves them to the right spot.
1 Like
Confirmed working for me.
Thanks for a superb product!