Incus admin init add member to the cluster using loop0 instead of a subvol (btrfs driver)

Incus
1

hello, I am trying to create a cluster where the first node was bootstrapped via incus admin init it successfully managed to create the local pool as a btrfs subvol, but while I try to join the other nodes following the same procedure they are create as a loopback device instead.

first node

df -h -T |grep btrf
/dev/sda3      btrfs     238G  3.0G  233G   2% /
/dev/sda3      btrfs     238G  3.0G  233G   2% /var/lib/incus/storage-pools/local

remaining nodes

df -hT |grep btr
/dev/sda3      btrfs     238G  3.0G  233G   2% /
/dev/loop0     btrfs      30G  5.8M   30G   1% /var/lib/incus/storage-pools/local

what answer should I give to this question in order to make it a btrfs subvol, I have tried the full path, relative, and soon but with no success ?

Choose "source" property for storage pool "local":

I am using the lts version.

2

Welcome!

When you setup a cluster, you can either use for a distributed filesystem (like Ceph) or traditional non-distributed filesystems. In the latter case, each node requires some configuration for a default storage pool, even if you end up using the storage pool of a specific node.

But in your case, do all nodes have their own independent btrfs subvol and someone you are not prompted to use those? That would be the case if sudo incus admin init does not prompt for a btrfs volume. Here you would need to install the btrfs package that provides the client tool for your Linux distribution.

3

Passing /var/lib/incus/storage-pools/local as the source in this case should do what you want, that’s effectively what we do on the initial server when you chose to use a subvol on the current disk.

4

Hello folks,

I solved my issue by recreating the pool, but indeed, passing the full path works as well.

Tks everyone