Proxmox VE local ZFS's + Cluster

[Kasproso]

Hello all

Last weeks I'm playing a little bit with Proxmox VE and I'm starting to know it and it's no a good way to stay maybe for a longer. But i'm stuck with something.
Previously I was using LVM to store VM's but when I try to create a snapshot it doesn't work. I need to use ZFS instead. Previously I use bare XEN and KVM and only LVM works for snapshots. So to do that, I've destroy my volumes and create ZFS's. I forgot to mention, that I have two nodes for the moment because I want to test how clustering works.

So I've delete LVM's on both nodes, create ZFS's, destroy Cluster and all configurations ans I've start almost from beginning. Now i have create storage on both nodes from Datacenter>storage>Add (ZFS). all looks fine. Now I'm creating new cluster, I have two nodes node1 and node2. Node1 is main one, on on node2 I'm joining to the cluster configured on node1. After that on node1 cluster I can see both nodes, which is good, however ZFS storage from node2 disappear and on it's place I see only ZFS from node1 with question mark, and I can not access to it.

What I'm doing wrong. Is there any documentation how to do that? Or does it possible to have such configuration at all?

Something like
Cluster:
Node1
-----ZFS-NODE1 (1,5T)
Node2
-----ZFS-NODE2 (4T)

And replication between them?

Thanks

 

[dcsapak]

Previously I was using LVM to store VM's but when I try to create a snapshot it doesn't work. I need to use ZFS instead.

you could also have used lvm-thin instead

So I've delete LVM's on both nodes, create ZFS's, destroy Cluster and all configurations ans I've start almost from beginning. Now i have create storage on both nodes from Datacenter>storage>Add (ZFS). all looks fine. Now I'm creating new cluster, I have two nodes node1 and node2. Node1 is main one, on on node2 I'm joining to the cluster configured on node1. After that on node1 cluster I can see both nodes, which is good, however ZFS storage from node2 disappear and on it's place I see only ZFS from node1 with question mark, and I can not access to it.

on a cluster join, the to joined node discards its config and takes up the one from the cluster so this is expected

what would have worked better if both storages (zpool names and storageids) would be the same, replication should work then
so :

node1: create zfs storage with id "FOO" on zpool "BAR"
on node node2: create zfs storage with id "FOO" on zpool "BAR"

then join