-
Notifications
You must be signed in to change notification settings - Fork 73
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
UEFI boot entry referencing /boot/efi/EFI/almalinux/shimx64.efi doesn't get created #31
Comments
Hello! However are you sure the systems won't come up again after reboot? Is updating boot entries mandatory? The only thing we had to do is to place grub configuration to new location: almalinux-deploy/almalinux-deploy.sh Line 325 in c545a5b
In Oracle's centos2ol conversion script we can see quite similar lines: So I'm not sure that current |
Hi Andrew, I'm a little bit short of time today for a detailed response because of Easter and I'm driving over to my family (Happy Easter! BTW). But to be clear, without adjusting the boot record my UEFI test systems doesn't come up running |
Sorry for coming back so late, the week has just been very busy. To reproduce, I just did again a fresh installation of CentOS 8.3 (PXE boot Kickstart from a as of today up-to-date local mirror) on real UEFI hardware (no VM, it's a Fujitsu Esprimo Q920). Hardware details:
This is the content of
As you stated,
Nevertheless, it is
As the development of
After the migration the situation in
Again
But looking at the EFI entries it all doesn't really matter, because the active record is still the CentOS one (with the files removed) and no new entry for AlmaLinux:
So I am really, really wondering how this should work. During an initial installation of the operating system, it's Anaconda which takes care of creating relevant boot entries, but this isn't the case with the migration script. There is also no RPM package which handles this as part of some embedded script: The only two RPMs which are doing something with
That's the reason why I think we are missing something really important here and I would be very surprised it this works for anybody on UEFI hardware. If I now reboot the machine, of course it doesn't come up again (no valid boot loader found). I actually did, it's just a test box. |
Thanks for your work on this one @cschug ; we appear to be having a similar EFI issue on a PROD VM server. Unfortunately, the server was mistaking created with EFI enabled instead of legacy BIOS mode, due to a VMware default which was not picked up by our team. The VM is running CloudLinux release 8.3 (Pavel Belyayev), but is actually a converted AlmaLinux 8.3 install, running WHM+cPanel. Everything works fine from a WHM+cPanel perspective; however, the /boot/efi partition is similarly messy:
As you can see:
We are now in the precarious situation that we can not reboot the VM, as grub will fail to find the images to boot. Before finding this bug report we have attempted several fixes using grub2-install and grub2-mkconfig, to no effect. We have a running VM snapshot we can roll back to when testing, but that is not really sufficient for a PROD workload. Any assistance on this issue would be greatly appreciated. |
Hi - on my first test system of using the deployment script, I too fell into this issue: unbootable system. Most Dell systems should have this issue since the BIOS creates bootmanager entries for CentOS referencing \EFI\centos\shimx64.efi. Since those files are moved/removed and efibootmgr is not used to create a new boot path, the system failed to boot (failed to PXE booting). I was able to visit the BIOS Setup and create a new 'alma' boot entry and it was easy to follow the path to EFI\almalinux\shimx64.efi. But surely, this should be automated with the deployment script. Next time I will also log all the output, but that too could be something the script could do itself. I'll dive into this more as time permits. |
Nice Bug. I stumbled over this today. Thanks for pointing out the problem. Changed boot path in Dell Bios to EFI\almalinux\shimx64.efi and machine ist starting up again. |
I had switched to Almalinux from Centos 8 before with no issues before at my work. The environment was VM with no EFI boot. Sadly, at home, I have also a NUC with a UEFI boot. So, when I used again the script after I rebooted, the server stuck on the boot. Here are my steps if somebody would have the same issue.
With efibootmgr -v you can see all the entries and disable or delete which references the old Centos installation. |
Update for almalinux-deploy.sh was just pushed and now it creates boot entry for AlmaLinux. |
Seems to have solved the issues I was having with CentOS v8 UEFI systems. Thanks. |
On UEFI hardware as of rev.
c545a5b
, almalinux-deploy.sh doesn't handle the changed boot loader path/boot/efi/EFI/almalinux/
. When migrating from CentOS to AlmaLinux this wasn't an issue up to AlmaLinux 8.3-rc where the path was still/boot/efi/EFI/centos/
but it is now with the final release as no new UEFI boot entry gets created. If one reboots after runningalmalinux-deploy.sh
, UEFI systems won't come up again.it's a bit late today for me to come up with a fully tested PR, but the basic idea for UEFI systems would be something like
The text was updated successfully, but these errors were encountered: