Can I Reinstall but retain old ZFS storage?

[Pandamonium]

Hi All

Homelab user here. I've spent the last few weeks getting familiar with Proxmox and moving all my old lab VMWare to it. Very happy so far, but...

I now want to reinstall on an old SSD as a test of rebuilding with a view to maybe buying some new SSDs later as everything I have at the moment is spinning disks. Here is my current setup...

• Proxmox OS and local-zfs on mirrored pair 3TB drives.
• Data zfs on mirrored 12TB drives.
• Empty zfs on mirrored 2TB drives.

I would like to move all the LXC and VM drives from local-zfs to the empty zfs leaving the current 3TB pair as only the OS, pull this pair of drives and replace them with the SSD (to be a mirrored pair in the future hopefully).

I have a some of questions before I try...

1. This server is part of a cluster (no HA or anything). I assume I will have to remove it and re add it when it is rebuilt?
2. I assume I would have to do a fresh install on the SSD but the question is will I then be able to add in my existing other zfs pools back to the new OS? While I do have backups, given the size of the data, I'd much rather just reattach the old storage if possible as I won't be touching those disks. Although it is part of a cluster, I don't have disk space to move them to the other server so I would like to leave the existing volumes where they are if possible.
3. Can I backup some config from the current PVE which will allow me to bring back the LXC and VM configs? Again, I have backups but I'm hoping there might be an easier way.

Any suggestions would be great please.

TIA.

 

[Johannes S]

You could move your vm/lxcs storage data to the empty zfs disc:

Then you would save your existing config in /etc/pve with tar or rsync and restore it after reinstall.

In any case:
- First test this with a VM with PVE first. So install a VM with PVE, replicate your current setup (of course with smaller discs but to reflect your existing storage layout) create some containers and (if you have nested virtualization support) vms and try out the move. The snapshot feature for a vm is quite handy, since you can move in time if something wents wrong.
- Create backups from /etc/pve, all vms and lxcs first (preferable to an external storage e.G. a NAS share or a removable USB drive) so you can still restore them if your resetup doesn't work out
- Test restore of the vm/lxc backups
- The backup/restore procedure can (and should) also be simulated in a PVE VM

Depending on your skill level and potential problems it might be easier to just do backup everything, reinstall, restore everything.

HTH