Ubuntu Cinnamon and Xubuntu FTBFS due to failure in diversion process

Bug #2054341 reported by Simon Quigley
16
This bug affects 1 person
Affects Status Importance Assigned to Milestone
livecd-rootfs (Ubuntu)
New
Undecided
Unassigned

Bug Description

https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/noble/ubuntucinnamon/
https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/noble/xubuntu/

Here is a shortened copy of my summary from the Ubuntu Flavors Matrix channel after the rubber-ducking session today:

Ubuntu Budgie, Edubuntu, and Ubuntu Studio Just Work™️. Why do they just work?

From a technical perspective, here is how far I've been able to go...

 1. First, I diagnosed what was causing the issue in the first place... if Plymouth somehow fails to configure, which it did in this case, every package depending on Plymouth will also fail to configure. Therefore, there is a large list at the bottom, but in reality, it's all held up due to plymouth. (apt should really have better UX around this.)
 2. So, the problem is this: /usr/sbin/update-initramfs: 3: exec: update-initramfs.REAL: not found - now, I thought, did Plymouth change recently? I mean, not really. The version of plymouth used on the ISO is the same version present in successful build logs. Which means that update-initramfs.REAL used to exist but now doesn't.
3. I found the code which does this in live-build/functions, at line 1141. Originally, I thought this is an issue of the function not being ran. After all, line 151 in live-build/lb_chroot_layered does actually specify that the diversion is to be done iff this is the "root pass." That thought is actually wrong, because this line is present in the failed build log: update-initramfs: diverted by livecd-rootfs (will be called later)

Which tells me that Some Magic™️ ✨ exists only for Ubuntu Studio, Ubuntu Budgie, and Edubuntu to retain that diversion, because suddenly it's being removed early.

Simon Quigley (tsimonq2)
tags: added: rls-nn-incoming
Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

I think what is going on here is that plymouth fails to configure when the diversion is in plance if initramfs-tools is not installed (well... unpacked, proabably). xubuntu-desktop-minimal contains plymouth but not initramfs-tools, so creating a layer with just that task in fails. This is a new failure because before there was not a layer that had just the minimal stuff in.

budgie-desktop-minimal and ubuntu-desktop-minimal have initramfs-tools in (mostly by accident I think). There is a good chance these only work because 'i' sorts before 'p'...

Edubuntu and Studio don't have a minimal seed / task.

I'm not sure what the best fix is really. We could disable the diversion (I think it's only an optimization?). We could fix the replacement update-initramfs script to be a noop if it can't find the .REAL binary. Maybe that's a better idea.

Revision history for this message
Julian Andres Klode (juliank) wrote :

> We could fix the replacement update-initramfs script to be a noop if it can't find the .REAL binary. Maybe that's a better idea.

Sounds like the best cause of action for me.

tags: added: foundations-todo
removed: rls-nn-incoming
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.