I’ve been attempting to update from XCP-ng 8.2.1 (GPLv2) to 8.3. However, after booting to the ISO (via IPMI), it says “Only product installations that cannot be upgraded have been detected. Continuing will result in a clean installation, all existing configuration will be lost.”
Any idea why this is happening or if there is a path forward that would not lose my existing configuration and VMs?
I have only done the 8.3 beta to 8.3 release and it was fine. Going to be a while before I try to upgrade my 8.2 LTS.
Only thing I can suggest is to download the latest 8.2 installer ISO, and boot from that to see if it upgrades anything, and even then do a sudo yum update after to make sure you catch everything.
Thanks for the suggestion, @Greg_E. I’ve tried downloading, mounting and rebooting to the older (8.2) iso as well and am getting the same behavior where it doesn’t seem to think it can upgrade and is warning it will wipe everything. Maybe I should try building a bootable USB stick instead of the IPMI ISO mount.
That’s possible, maybe USB will work better.
Are you trying this on the pool master? And did you check to make sure that a yum update didn’t show anything that needs to be done?
All my IPMI/iLO are so slow that I get frustrated and just “burn” a USB drive then walk back to my office and pick up the process through the remote console. At least for my lab. I’m not moving production over for a while, no real improvements that I need for my production system right now.
If a USB and starting with the pool master doesn’t get this working, I’d say you are probably going to need to ask this on the XCP-NG forums.
Not an exact situation, but might help.
Thanks again! I read through the article you linked and it gave me an idea to try booting from BIOS instead of UEFI and that worked totally fine. It did warn me about BIOS booting being deprecated and it might be an issue for future updates, but it got me past the detection issue and I’ve now completed the upgrade. Fortunately my IPMI/iLO ran at a reasonable speed and was easier than trying to find an old USB stick somewhere.
I think I’ll plan to setup another XCP host beyond this single-host master “pool” so that I can move VMs around and eventually reimage this as a clean install with UEFI but I think this is going to work great for now. Thanks again for all your suggestions, @Greg_E!
Currently I’m waiting for it to start a bunch of my VMs–for the last 20 minutes or so. Seems like it’s going through a number of tasks that I hope are one-time only updates (lots of Xapi#getResource /rrd_updates (on xcpng) and some host.call_plugin and SR.scan’s and other tasks).
XO Lite looks cool. Finally dark mode! It’s still called “lite” for a reason. My other XO VMs are working fine.
That’s one I wouldn’t have been able to test, my lab is old enough that it doesn’t support UEFI.
I did get an official statement about the requirement for UEFI… Xen Server is going to force this, but XCP-NG hasn’t yet decided if they are going to follow this requirement. Apparently they are still talking about what/how they are going to handle this. The only real con is that older hardware can no longer be used for your labs, I’d hope no one is running stuff this old in production (even though my lab has been really good).
If a Google search brought you here because of the “Can’t be upgraded error” I’ll just offer some info. You can see the more specific reason for not being able to upgrade in the install log. To get to that, press ALT-F2 to switch to another console and use the command prompt to inspect /tmp/install-log
The issue in your case is that you can’t use the upgrade process to switch between UEFI and BIOS installations. The reason for that is technical. While theoretically possible it would make the process more complicated because the two types require different disk partition layouts and safely changing the layout while preserving settings could be complicated.
In my case, I had removed a test host from my pool in order to do a test upgrade on that host before trying anything else. Removing a host from the pool resets it and there are no initial setting for the upgrader to work with, so I needed to reboot the test host into the old XCP-ng install at least once after it had been removed from the pool to initialize its settings.
P.S. I didn’t end up upgrading at all at this time because I still use XCP-NG Center in cases where XO is being weird or hasn’t implemented something. XCP-NG Center no longer works with 8.3. And XO-Lite is really incomplete.
Thanks for the tip. This was the solution for me. I was using the virtual drive on the JetKVM, it uses UEFI so the upgrade won’t work. I ended up creating a bootable USB drive starting the upgrade using bios and it worked for me. One thing to point though, based on the message it seems at somepoint in the future we won’t be able to use bios and UEFI will be the be the only option. We’ll have to do a clean install. I’ll cross that bridge when I get to it. Thanks again.
