Cluster - possible PVE 5.x and PVE 6.x ?

[kappclark]

I think the answer is "NO!" to the question:

Can I cluster a PVE 5.x host with a PVE 6.x host ? If no, does the second box need to be the exact 5.x release of proxmox ??

I have i host currently running 5.x Good to know befor I build my next box

 

[wolfgang]

Hi,

Proxmox VE 5 and Proxmox VE 6 are not compatible in case of clustering.
PVE 5 use corosync protocol version 2 and PVE 6 corosync Version 3.

If no, does the second box need to be the exact 5.x release of proxmox ??

We so only test clusters with the same version.
So if you mix version you use an untested setup.

 

[kappclark]

Thank you for the reply --
I have 5.3.8, and now see the download of 5.4 -- will these be compatible for cluster ? Or do I need to update the 5.3.8 to 5.4 ??

 

[Glowsome]

As stated above Clustering on PVE5 is not compatible with PVE6 due to the way in which corosync version 2 differs from 3, however for migration purposes there is a way :

My approach would be :

1. moving towards the latest 5.4 version.
2. check compatibility on all your nodes for migrating towards PVE6 ( run pve5to6 ) and fix found issues ( important !)
3. Then migrate (on all nodes) corosync2 => corosync3 ( as described in https://pve.proxmox.com/wiki/Upgrade_from_5.x_to_6.0#Cluster:_always_upgrade_to_Corosync_3_first
4. test cluster operations, remember this is a 'migration-state' and not officially supported (as stated above)

As i have a 4-node cluster fully migrated to PVE6 i have not tested this partial PVE5/PVE6 state, nor have i remained in this situation.
So i cannot speak out of experience, but just giving my thoughts about it.
I migrated thru all nodes when i was convinced the upgrade went ok.

 

[kappclark]

wow - thank you for the advice... - I will review the wiki ..

I tried to upgrade the 5.3.8 to 5.4 and was not initially successful. But then, I did some more searching and found this link which explained updating the package lists to no-subscription -- a few edits with nano opened up the update engines..looks like I have 5.4-2 running now..pve manager shows as 5.4-13 which is what I see on the web management page..

root@proxmox:/# pveversion --verbose

proxmox-ve: 5.4-2 (running kernel: 4.15.18-10-pve)
pve-manager: 5.4-13 (running version: 5.4-13/aee6f0ec)
pve-kernel-4.15: 5.4-9
pve-kernel-4.15.18-21-pve: 4.15.18-48
pve-kernel-4.15.18-10-pve: 4.15.18-32

I am planning to build a test host with 5.4.1 today...I will let the forum know what happens when I try to make a cluster between the 2 pve5 boxes..

 

[Belokan]

Hello guys,

Sorry for digging up this old subject but my question is 100% related to it

I've not been really watchful with my pve@home cluster for some time and only took care of the guest's maintenances. But now I've started to upgrade my 3 nodes pve/ceph cluster to 6.x.

So far, all 3 nodes are now running:

- pve 5.4-15
- corosync 3.0.4
- ceph 12.2.13
- kernel 4.15.18

Except 2 VMs OpenMediaVault attached to a specific node because of storage pass-through, I'd like to keep some infrastructure VMs as much running as possible (DNS, DHCP, gateway, ...).
So in this condition, can I upgrade my nodes one after the other and switch VMs (pve5 -> pve6) to the newly updated before updated another one ?

If yes, must I update ceph together with pve on each node or only once all nodes are running pve 6.x ?

Thanks a lot in advance !