Suggestions for vm/containers while performing ceph upgrade to squid

[Heracleos]

Hello, everyone.
I am planning a Ceph upgrade from reef to squid, of my proxmox ve 8.4 cluster.
Clearly, I will use the official way as described here:
https://pve.proxmox.com/wiki/Ceph_Reef_to_Squid

However, it is not clear to me how I should manage the vm's and containers while doing the upgrade.
To wit:
1) should I turn off the vms and containers on all nodes while I upgrade ?
2) should I turn off or migrate the vms/containers from the node where I am upgrading ?
3) can I leave the vms running while I am upgrading ?

Thanks all for any advice.

 

[groque]

Hello,

I recently went through a similar upgrade (Reef → Squid) on a Proxmox VE 8.x cluster, so here’s some guidance based on both the official Proxmox recommendations and my experience:

1. Should I turn off VMs and containers on all nodes while upgrading?
   No, you do not need to shut everything down all your VMs/LXCs cluster-wide.
2. Should I turn off or migrate VMs/containers from the node being upgraded?
   It's not strictly necessary. Ceph is designed for rolling upgrades, meaning you can upgrade one node at a time while the rest of the cluster remains operational. If your cluster is healthy (or HEALTH_WARN only due to noout), the upgrade shouldn't affect running VMs and LXCs. However, I prefer migrating VMs and containers as a precaution to ensure workloads remain unaffected in case any unexpected issues happen during the upgrade process.
3. Can I leave VMs running while upgrading?
   Yes, VMs and containers can stay online on other nodes during the upgrade. Ceph remains operational throughout if you follow the recommended rolling upgrade procedure and avoid taking down multiple OSDs or MONs at once.