Add new 6.2 node to a 6.0 cluster

[Joel LaCalamita]

I have a cluster (6 nodes) at version 6.0. Is it safe to just add a new version 6.2 node?

With addition of a new node I'll have more capacity (RAM) to live migrate VMs off existing nodes, update them and reboot for new kernel. RAM on existing nodes is highly utilized +90% making live migration a challenge.

 

[t.lamprecht]

I have a cluster (6 nodes) at version 6.0. Is it safe to just add a new version 6.2 node?

Yes, the join process of nodes with the same major release is normally compatible.

If you use the HA stack and want to be on the safe side you could disable HA during the join process, avoiding that a hiccup takes out some nodes.

 

[Joel LaCalamita]

I have HA configured and enabled only for two VMs at this time.

Do you mean "Stop" the pve-ha-crm & pve-ha-lrm daemon on the new node before joining it to the cluster and starting them after it has successfully been joined? If so is there a recommended sequence for restart?

 

[t.lamprecht]

Yes, it shouldn't be required but if you want to play it more safe, then yes.

For procedure:
The important thing is that you stop the LRM first, so that there's still a manager around which they can talk to on shutdown.
After all HA services are in the "frozen" state, you can stop all CRM services on all nodes too.

After the join you'd start in reverse, first start CRMs then start the LRMs.

 

[Fiko]

Hi, Same situation here, I had 6 nodes with V6 and added 3 nodes with V6.2, I can't live migrate VM from older version nodes to the new ones as it give me an error (not the same version). All of the vm are HA using ceph as main storage.

As I can't stop the VM's as they need to run 24/7 and can't live migrate, is their a way to force live migrate.

 

[Dan S]

Hi.

Joel, did you face the same problem as Fiko?

We are going to add new nodes to a 6.0 cluster allow enough safe resources in order to move machines from the 6.0 servers and upgrade them to 6.2 also.

But Fiko says that it didn't allow him to have live migration from 6.0 nodes to 6.2 nodes into the same cluster. Could you confirm that Fiko? Was it solved?

Thanks a lot.

Daniel.

 

[Fiko]

Hi.

Joel, did you face the same problem as Fiko?

We are going to add new nodes to a 6.0 cluster allow enough safe resources in order to move machines from the 6.0 servers and upgrade them to 6.2 also.

But Fiko says that it didn't allow him to have live migration from 6.0 nodes to 6.2 nodes into the same cluster. Could you confirm that Fiko? Was it solved?

Thanks a lot.

Daniel.

I've added 6 more nodes, so, moved machines from old version nodes to new ones and upgraded, etc...

 

[Dan S]

I've added 6 more nodes, so, moved machines from old version nodes to new ones and upgraded, etc...

But where you able to live migrate machines from older proxmox (v6.0) to the newer ones (v6.2)?

I understood that you face a problem with that and couldn't do it live, and needed to stop the VM, move it and start it in the new server.

Thanks a lot.

Daniel.

 

[Fiko]

But where you able to live migrate machines from older proxmox (v6.0) to the newer ones (v6.2)?

I understood that you face a problem with that and couldn't do it live, and needed to stop the VM, move it and start it in the new server.

Thanks a lot.

Daniel.

Never stopped the vm as we use Ceph for shared/replicated storage. Just live migrate the VM.