Reinstall PX on internal SSD without touching RAID10

[Th0mas51]

Hello,

I have a basic question, I've been searching on Google and on the forum, but I cannot find the answer.

My company is running a ProxMox servers, with the following settings :
- PX Version : 4.2
- installed on an internal 128GB SSD
- VMs are stored on 4 x internal 2TB disks configured as RAID10 (So 4TB of storage in total)

We would like to upgrade to PX 5, but we cannot migrate the VMs out of the RAID10 because it will create a too big downtime.

Is it possible to format the SSD and reinstall the new PX version on the SSD, without touching the RAID10 partition, and then re-declare the VMs that are on the RAID10 in the recently installed ProxMox OS ?

I was thinking that I just need to backup ProxMox config files, install the new version, reconfigure like before, and redeclare the VMs in the new ProxMox OS.

Anyone tried this operation already ?

Thanks

Thomas

 

[aPollO]

The safest way is by pyhsically disconncting the RAID disks before installing the new Proxmox version.
Normally you can choose a disk for the installation in the setup assistent. The other disks will not touched normally. Then you can redeclare you storages and import the backuped config files.


But i dont recommend to use the Beta Version for productive use.

 

[fortechitsolutions]

Hi, just to chime in.
- This is doable, unplug the raid disks first is a good suggestion to avoid 'oops' during the clean install
- to be extra safe, I would suggest you first run the day-or-two before hand, a no-downtime VM-dump-backup-everything-to-external-NFS-storage. Just in case things go to hell, it is nice to have full VM backups that are a few days old or less.

Then you have basic routine,
- backup config (ie, tarball the /etc/pve to /your/raid/data/storehere.tgz for example)
- unplug raid disks
- reboot, clean install proxmox 4.X probably for production use
- once done, re-attach raid disks. Make sure you have linux-raid support bits installed
- edit your fstab, create mount point, mount the raid, make sure you have visibility into the raid disk volume / ext4 filesystem (whatver FS type it is)
- then you can happily start to restore configs and make things -reexist from where they were not before.
- and do it carefully so you don't accidentally destroy your only good copies of the production VMs


Even safer, put aside the original 128gig SSD (heck they are cheap) pop in a new one, do clean install there. In case life goes sideways, you have an exit hatch (return to original SSD drive) and getting back into production as per your pre-upgrade config. With modestly no work and modest cost (price of SSD 128gig, not quite zero but pretty close)


Tim