Upgrade Proxmox Cluster from 8.4 to 9 – Can I migrate VMs back to the upgraded node before finishing all nodes?

torbho

torbho

New Member
Proxmox Subscriber
May 8, 2025
15
2
3
Hi everyone,

I’m running a 3-node Proxmox VE HA-cluster (currently 8.4.12) and I’m planning to upgrade to version 9.

My question is:
  • After upgrading the first node to Proxmox 9, is it safe to migrate VMs back to this upgraded node while the other two nodes are still running 8.4?
  • Or do I have to shut down the VMs until all nodes are on the same major version?
I’d like to minimize downtime and I want to make sure I don’t run into cluster or migration issues due to mixed versions.

I couldn't find anything to this specific scenario in the documentation.
Thanks for clarifying!
 
Yes Proxmox makes sure that you can always migrate to a newer node (but not back from the new node to an older one). Livemigration is no problem, have fun!
 
Thanks a lot for the clarification regarding migration direction (8.4 → 9 is fine, but not back). That helps a lot!

One more thing I came across: I noticed that Proxmox 8.4 and 9 ship with different Ceph versions. My virtual machines do use Ceph.
Does this cause any issues in a mixed cluster during the upgrade phase?
 
jsterr said:
Yes Proxmox makes sure that you can always migrate to a newer node (but not back from the new node to an older one). Livemigration is no problem, have fun!
Click to expand...
Thank you!

pvps1 said:
follow the upgrade instructions strictly (!)
ceph upgrade is always mentioned as a special step.

worked like charm for the last decade IF you follow official howto
Click to expand...
Thank you, too.

I will try the upgrade in some days. *fingers crossed*
 
Just for clarification: the upgrade from the Proxmox VE Cluster 8.4 went almost perfectly. There was only a small issue with Ceph — it didn’t synchronize immediately after upgrading the last node. After about a minute, the problem resolved itself, but the VMs were offline during that time.
 
  • Like
Reactions: Johannes S
torbho said:
Just for clarification: the upgrade from the Proxmox VE Cluster 8.4 went almost perfectly. There was only a small issue with Ceph — it didn’t synchronize immediately after upgrading the last node. After about a minute, the problem resolved itself, but the VMs were offline during that time.
Click to expand...
Im thinking about doing this exact thing, how did you go about doing it ? Anything not in the official how-to to? I have a 3-node cluster and am looking to install 9.0 on one node, then upgrade from 8.4.14 on the other two after migrating the services to the new node (9.0). Any suprises besides ceph that you can recall? Or advice? : )
 
I followed the official guide exactly as described. Here’s how I handled it step by step:

I first migrated all VMs from one node to another within the cluster. Then I upgraded that empty node to Proxmox 9. After confirming everything worked fine, I moved the VMs to the third node and upgraded the second one. Once two nodes were already on 9, I distributed the VMs back across those updated nodes and finally upgraded the last one.

A few notes from my experience:
  • Make sure you’re accessing the web interface from a node that’s already finished upgrading, not from the one currently being updated.
  • Ideally, have IPMI or some other form of out-of-band management available, in case SSH or web access becomes unavailable during the upgrade.
  • And very important: if you’re using Ceph, first make sure to upgrade from Reef to Squid as described in the official documentation — that step is crucial and also mentioned in the guide.
Beside the ceph-problem, I didn’t run into any unexpected issues — as long as you follow the steps carefully, it’s pretty straightforward.
 
Last edited:
You must log in or register to reply here.
Top Bottom