7

I need to upgrade the OS of my ZFS disk server from CentOS 7 to Rocky Linux 9. I am currently running ZFS 2.0.7, and if I'm correct Rocky Linux 9 uses 2.3. My tentative plan is to use the AlmaLinux Elevate scripts to perform the migration to Rocky Linux 8 and then upgrade to Rocky Linux 9, as explained here. The OS resides in a dedicate SSD.

Since the server size is quite big (24 x 16 TB disks in ZRAID2 configuration) I have no backup possibility. Data loss is not an option, so I'm worried about possible troubles, e.g.:

  1. is there any chance that upgrading the OS would result in pool corruption / data loss?
  2. should I expect issues due to the upgrade of the ZFS version?
  3. is the linked upgrade procedure robust? Could it be more safe to simply do a fresh install of Rocky 9 and reconfigure everything?

Thanks in advance for any advice.

1 Answers1

7

If there is not much to configure, I'd perform a fresh install. But your plan looks sensible, and if there is much configuration (and it wasn't automated), the upgrade path is to go. I went it once (I was upgrading from Centos 7 to then-recent AlmaLinux, but that shouldn't matter), with success.

The backup is mandatory. You should expect you'll screw up a few times, so the ability to restart from scratch is a must. However, you don't need to backup the datastore; only the system. I believe you can afford backing up that, considering it's residing on the separate storage.

Physically detach the pool, backup the system, and then either reinstall or upgrade while pool is detached. Newer ZFS is able to import the older version pool, so once you determined you have everything to your liking, you'll attach the pool back and then upgrade it (also look here).