How to reproduce:
[FAILED] Failed to start Authorization Manager
All attempts to boot in rescue mode on non-graphical mode result into fail and we got XFS errors:
<img alt="xfs_error.png" src="/centos-sig-alt-images/spin-bugs/issue/raw/files/dd6c2f786352ddb024111623ab4165231f2874a2b4690e3b07713a284885d60b-xfs_error.png" />
This bug doesn't reproducible on non-Live iso: CentOS-Stream-9-20240122.0-x86_64-boot.iso.
CentOS-Stream-9-20240122.0-x86_64-boot.iso
Metadata Update from @ngompa: - Issue tagged with: GNOME, c9s
Were you able to "reboot" after turning on auto-login? For me it hangs, and I have to force it off. But once, it did show an XFS "Corruption of in-memory data" before it hung. Very strange. I don't see how it could be a result of turn on auto-login, but ... anyway ... investigating more.
OK, it has nothing to do with the auto-login. Just rebooting after doing the updates with GNOME-Software will do it. Still investigating.
It appears to be the GNOME-Software update that is causing the problem. What exactly, I don't know yet.
Using the KDE Live image. Not a problem updating with either the command line or the graphical tool (Discover). Using the Gnome Live image, not a problem using dnf by hand.
But if I use GNOME-Software to update, it seems to be messing with the filesystem in some way. I'm wondering if it's because it downloads, reboots, updates almost everything, reboots, and then finishes off stuff. I'm wondering if it really should be running the scripts before the reboot.
I'm not the best at diagnosing the GNOME-Software package. So if others want to jump in here, that would be good.
It appears to be the GNOME-Software update that is causing the problem
What is interesting that i can't reproduce this bug in non-Live CentOS image (CentOS-Stream-9-20240122.0-x86_64-boot.iso). I also installed updates via GNOME Software. I'll try to double check this.
Could this caused by some difference in partitioning in Live image? One thing i noticed is that LVM volume naming differs in Live image: cs_localhost--live-root. Have no idea how this could result into XFS corruption but who know. Also maybe we need to investigate SELinux here.
cs_localhost--live-root
We've got our new quarterly images out. I'm going to test this again with these new images once we have some CentOS Stream 9 updates out.
I think we discovered the problem. We had unreleased packages on our images due to the buildroot being in our repo's, and weak rpm dependencies. See this issue: https://pagure.io/centos-sig-alt-images/sig/issue/2
I believe this has been fixed with this quarters images, as noted in the other issue. I will wait a few weeks so that we have some updates, and test again to see if we are fixed.
Fixed. You need to start with a newer image (202407+), but it is fixed in the newer images.
Metadata Update from @tdawson: - Issue status updated to: Closed (was: Open)
Log in to comment on this ticket.