Unknown disks on one node after joining the cluster

[freak12techno]

I have 2 proxmox servers (proxmox-1, proxmox-3, both PVE 7.3). After creating a cluster on proxmox-1 and joining it on proxmox-3, I see additional disks on proxmox-3 host in the UI, with the same names as the disks on proxmox-1, but in status "unknown":

/etc/pve/storage.cfg on both nodes (looks identical):
dir: local path /var/lib/vz content vztmpl,backup,iso zfspool: local-zfs pool rpool/data content images,rootdir sparse 1 lvm: disk1 vgname disk1 content images,rootdir shared 1 lvm: disk2 vgname disk2 content images,rootdir shared 1 lvm: disk3 vgname disk3 content images,rootdir shared 1 lvm: disk4 vgname disk4 content rootdir,images shared 1 lvm: disk5 vgname disk5 content images,rootdir nodes proxmox-3 shared 1 lvm: disk6 vgname disk6 content rootdir,images nodes proxmox-3 shared 1 lvm: disk7 vgname disk7 content rootdir,images nodes proxmox-3 shared 1 lvm: disk8 vgname disk8 content rootdir,images nodes proxmox-3 shared 1

How can I resolve it so they will be displayed properly?

 

[fabian]

you need to restrict local storages that are not available on all nodes (can be done on the GUI -> Datacenter -> Storage -> Edit -> "Nodes")

 

[freak12techno]

you need to restrict local storages that are not available on all nodes (can be done on the GUI -> Datacenter -> Storage -> Edit -> "Nodes")

I already set all of the disks as shared (as I want them to be accessible on all servers), didn't help, and all of them were not shared before, which was leading to the same error.

UPD: after updating all the disks on proxmox-1 to be available to proxmox-1 only and same for proxmox-3 it seems to be fixed. It doesn't like when I enable the disk on multiple nodes though, but it seems to do the trick for now.

 

[fabian]

well, "shared" means that you tell PVE that this storage is available and has the same contents on all nodes (i.e., because it's an export from a file-server, or something like Ceph/Gluster/DRBD/..), not that PVE will somehow magically share it for you

 

[Pyromancer]

you need to restrict local storages that are not available on all nodes (can be done on the GUI -> Datacenter -> Storage -> Edit -> "Nodes")

Just wanted to note this answer has solved a problem I was having, where the one machine in a 3 node cluster that doesn't have a /tank2 was showing tank2, but unknown status, which in turn was causing any attempt to migrate VMs off that node to fail with "zfs error: cannot open 'tank2': no such pool", even though the VM in question only referenced tank1.

Datacentre -> Storage -> Highlight tank2 -> Edit -> Nodes

Bit scary at first as I thought it was only going to allow "All Nodes" or a single one, but it lets you pick more than one host before you submit it. So tank2 now exists on nodes 1 and 2 but not 3, and I've just successfully migrated the first VM from 3 to 2, in preparation for the upgrade.