This message was deleted.
# k3s
a
This message was deleted.
c
the local-path-provisioners is for just that, local paths. It doesn’t expect anything in there to be replicated across nodes, even if it is it won’t be aware of that.
If you have an existing NFS server that you’re using, setting up the external NFS provisioner is probably your best bet
n
I do not have nfs. It's ceph, provisioned by proxmox, with block storage made available inside each node.
c
ah. Then you might be able to use the Ceph CSI driver
There’s no real way to just tell kubernetes “this is the same shared filesystem on all the nodes”. You really need to go through the volume abstraction layer, and the best way to do that is to find a CSI driver that works with your storage backend.
n
I see. Alright, I'll look into doing it that way. Thanks!
Looks like the status is not in a great place for ceph. Alpha, with no targets listed for beta or GA - https://kubernetes.io/blog/2021/12/10/storage-in-tree-to-csi-migration-status-update/
c
That’s for migration, if you’d been using the in-tree volume driver and are trying to migrate to the CSI driver. The standalone CSI driver is quite mature. https://github.com/ceph/ceph-csi
n
This looks good. Thank you!