23.10.1 update and UEFI-VMs boot problems

[Whiskydrinker]

I've got two VMs with Debian 12 that I created under 23.10.0.1 and that are configured as UEFI machines. Both used to boot fine under that release.
Now I decided to upgrade to 23.10.1. And that prevents the machines from booting. All I get is the UEFI shell.

So for testing reasons I deleted one VM keeping its HDD. Then I recreated the VM under 23.10.1 attaching the HDD from the deleted one.
I still ended up getting the UEFI shell only.

My next step was rebooting the system with the release 23.10.0.1. And there the untouched and the recreated VM boot fine again.

Anybody having similar problems with the 23.10.1 update?
And what has changed with the update 23.10.1 regarding virtualization that could break things?

 

[beagle]

There is a workaround reported on another post:

To work around the VM issue:

1. Run FS0:\EFI\debian\grubx64.efi in the EFI shell (via a SPICE display, setting up a temporary one if needed) to boot.
2. Log in at the Debian TTY.
3. Follow Debian's instructions to install GRUB to the "EFI removable media path".

Again, I am not sure this is a TrueNAS regression... perhaps something changed in a recent Debian update. It's just weird I didn't need that workaround with Debian 12 + TrueNAS SCALE 23.10.0.1, but it is needed again with Debian 12 + TrueNAS SCALE 23.10.1.

 

[bcat]

Ah, fascinating. Thanks for confirming that rolling back to 23.10.0.1 resolved this issue! It sounds like it is a behavioral change in SCALE, then, and might be worth filing a bug about (though the workaround is straightforward for the time being).

 

[seanthegeek]

Someone filed a bug here.

https://ixsystems.atlassian.net/browse/NAS-125816

Personally, I'll stick to rolling back to 23.10.0.1. I have quite a few VMs, and I'd rather not have to apply a workaround on each one.

I see that the fix version is currently set to SCALE-24.04-BETA.1 (DragonFish). Does that mean this bug won't be fixed until the next major release? I hope not.

 

[seanthegeek]

I found two better workarounds thanks to the Debian Wiki.

Download the CD image of rEFInd and attach it to a CDROM as the first device of your VM.

This will boot your system normally by default within a few seconds.

You can either leave the system in this state, in case the TrueNAS team finds a solution that doesn’t require reinstalling the bootloader of each VM or boot the system this way in TrueNAS 23.10.1 and reinstall the bootloader so that the ISO workaround is no longer needed.

To reinstall the bootloader, run the following commands:

sudo apt-get install --reinstall grub-efi
sudo grub-install
sudo update-grub

Then shut down the VM, delete the CDROM device, and start the VM.

Personally, I’ve opted to go ahead and reinstall the bootloaders of all of my Debian/Ubuntu VMs this way and take new snapshots.