Replace all SSDs on a 3 nodes cluster

CarmineM74

Member
Mar 6, 2021
2
0
6
51
Hi everyone,
I have a 3 nodes cluster running on PVE 6.4 with a total of 24 SSDs with CEPH.
Considering that:
- the cluster can be brought to a total stop
- each node is more than capable to host all the machines
- new SSDs are bigger (from 960GB to 1.92TB)
- I'd highly prefer to not stress the old SSDs further

I am planning to do as follows:

1. Remove everything from HA and stop all machines
2. Backup all VMs and containers to the local Proxmox Backup server
3. Shut down the nodes
4. Replace all the SSDs on one node and start it
5. Recreate all the OSDs
6. Restore backups on the running node
7. Repeat from step 4 through 6 for the remaining nodes

Would that be a good plan?

Thanks in advance for your suggestions and help
 
Last edited:
PVE 6 has been out of support for a year. Maybe install a new cluster of PVE 8 on the new SSDs and restore the VMs from backup once it works? You'll have the PVE 6 SSDs as a fallback if you run into unexpected issues.
 
PVE 6 has been out of support for a year. Maybe install a new cluster of PVE 8 on the new SSDs and restore the VMs from backup once it works? You'll have the PVE 6 SSDs as a fallback if you run into unexpected issues.
First off, thanks for your prompt reply and advice.
It would be nice to switch to PVE 8 but I have to be absolutely certain that there are no quirks with certain windows VMs which are key to the company.
Switching to PVE 8 to find that Windows doesn't like something about it on monday morning would be catastrophic.