using zfs on existing proxmox install

[chudak]

Hello

I installed proxmox on a ssd drive and it looks like:




If I want to use this drive a zfs shared storage, what would be a recommended way?
TIA

 

[Dunuin]

First, ZFS is no shared storage. You either need to share folder manually yourself without the webUI using SMB/NFS.
If you mean ZFS replication between nodes, this too isn't a shared storage. It's local storage on all nodes that gets synced between them every minute (or longer). So every node got its own local copy.

And there is no easy way to convert a root filesystem from LVM to ZFS. You would have to back everything up, and start from scratch installing PVE again but using ZFS.

 

[chudak]

First, ZFS is no shared storage. You either need to share folder manually yourself without the webUI using SMB/NFS.
If you mean ZFS replication between nodes, this too isn't a shared storage. It's local storage on all nodes that gets synced between them every minute (or longer). So every node got its own local copy.

And there is no easy way to convert a root filesystem from LVM to ZFS. You would have to back everything up, and start from scratch installing PVE again but using ZFS.

I was referring to these steps https://www.youtube.com/watch?v=FZnVmt_DvUk&t=499s on 8 min 10 sec
So the trick is to use the same name.

But can what is described in the video be achieved with one disk?

 

[Dunuin]

But can what is described in the video be achieved with one disk?

What do you mean? One disk per node?

 

[chudak]

What do you mean? One disk per node?

Like I pasted above
If I have one ssd (nvme) drive and need to use it for pve installation and replication.

 

[Dunuin]

Like I pasted above
If I have one ssd (nvme) drive and need to use it for pve installation and replication.

I don' see why this shouldn't work. You could manually create a new dataset like zfs create rpool/replica on all nodes and then create a new ZFS storage on your first node pointing to that dataset and selecting all nodes for it.

 

[chudak]

I don' see why this shouldn't work. You could manually create a new dataset like zfs create rpool/replica on all nodes and then create a new ZFS storage on your first node pointing to that dataset and selecting all nodes for it.

Thank you for reply
I don’t feel too knowledgeable about zfs.
Considering maybe I can buy and add 250gb ssh for pve install only and then use 1tb drive for replication