Proxmox Upgrade and Filesystem Change

[rghiglianovich]

Hello all,

I am new here, I'm writing to get some hints ...


Currently, I have a cluster consisting of 2 nodes running Proxmox VE version 5.4.34 with LVM storage on an ext4 filesystem. I'm looking to upgrade to the latest Proxmox version and switch to ZFS as the filesystem.

I'm exploring the best way to accomplish this. Initially, I had considered upgrading from version 5 to 6 and then from 6 to 7, but this approach wouldn't allow for a filesystem change.

I've been considering a potential solution: migrating all VMs and containers to a single node, removing the now-empty node from the cluster, performing a fresh installation of the latest Proxmox version, reconnecting it to the cluster, and repeating the process on the other server.

What are your thoughts on this approach? Have any of you already undergone a similar upgrade? Are there any other methods that are better and/or safer?

Thank you for your help,

regards,

RIccardo

 

[LnxBil]

LVM storage on an ext4 filesystem

maybe the other way around?

I've been considering a potential solution: migrating all VMs and containers to a single node, removing the now-empty node from the cluster, performing a fresh installation of the latest Proxmox version, reconnecting it to the cluster, and repeating the process on the other server.

Could work and is a lot of manual work.

What are your thoughts on this approach? Have any of you already undergone a similar upgrade? Are there any other methods that are better and/or safer?

What about backup & restore? It's the easiest an non-error prone approach there is.

 

[leesteken]

Currently, I have a cluster consisting of 2 nodes running Proxmox VE version 5.4.34

Two-node clusters are always problematic without a third vote and it will probably stop working when you take one node down for an upgrade. Best to install a new cluster on new hardware with the latest version and restore from backups (and maybe test that before stopping any of your current nodes).

 

[rghiglianovich]

maybe the other way around?

ops yes, maybe

Could work and is a lot of manual work.


What about backup & restore? It's the easiest an non-error prone approach there is.

do you mean: backup , fresh install and then restore? but without moving the VMs, the services are stopped

 

[rghiglianovich]

Two-node clusters are always problematic without a third vote and it will probably stop working when you take one node down for an upgrade. Best to install a new cluster on new hardware with the latest version and restore from backups (and maybe test that before stopping any of your current nodes).

Certainly it seems to be the best solution, anyway I am out of additional hardware

 

[LnxBil]

do you mean: backup , fresh install and then restore? but without moving the VMs, the services are stopped

Yes, backup & restores is the easiest way to move VMs around if you do not have a shared storage.

 

[rghiglianovich]

My goal is : upgrade the cluster to latest version pve, maintaining the current hardware

this is why I planned to reinstall the single nodes, moving the up and down the VMs

but I read somewhere that all the nodes must have the same exact version: is it true? If so it is, all my plans are garbage...


thanks again,
RIc