[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#1102604: rescue-mode: Provide rootfs subvolume selection menu



Hi,

Martin-Éric Racine <martin-eric.racine@iki.fi> (2025-12-25):
> We're now 2 weeks away from the next point releases. Can we fix this?

Thanks for the reminder.

Diving back in, the topic was backporting features that were published
in trixie and supposed to be working, to bookworm.

There were/are two options:
 - A blanket ~deb12u1 of the trixie version. Outside the changes we
   would want to backport, I'm only seeing translation updates, so
   that could work… but I'm not really a fan of seeing so many changes
   for opu.
 - Cherry-picking (only) the changes we want, and +deb12u1 the
   bookworm version.

I prepared a branch locally that implements the latter. Then I built a
netinst ISO using d-i components found in current bookworm (including
the text and gtk initrds), but shipping the modified rescue-mode via a
“local” directory.

That seems to be sufficient, as src:rescue builds two binary packages,
rescue-check and rescue-mode. The former is used when debian-installer
is built, and if there were changes there to be tested, I'd need to
switch from the current bookworm's d-i to a local build (which is
something I could do, if that were necessary). The latter is loaded
from the installation image at runtime, and I could confirm the
version (+deb12u1) and also the extra lines about btrfs in
/var/lib/dpkg/info/rescue-mode.postinst.

I haven't run too many checks:
 - I broke a workstation of mine (mv /lib/systemd/systemd{,.cassé}),
   verified it wouldn't boot anymore, and I couldn't really fix it
   without fiddling with keyboard mappings.
 - I started the custom ISO in rescue mode, then assembled the RAID
   arrays in automatic mode, selected the right LV as the root
   filesystem, agreed to mount the ESP, then got a shell in there,
   reverted the breakage, exited that shell.
 - I opened another shell in the d-i's context, verified the
   rescue-mode version and code were what I expected.

So as far as I know, the cherry-picks didn't break rescue-mode
horribly. That being said, I'm not volunteering to install some
btrfs-based systems to verify this ISO knows how to deal with them.

Netinst for amd64:
  https://people.debian.org/~kibi/bookworm+btrfs/

Please test and report back!


For reference, the git branch:
  https://salsa.debian.org/installer-team/rescue/-/commits/bookworm+btrfs


Cheers,
-- 
Cyril Brulebois (kibi@debian.org)            <https://debamax.com/>
D-I release manager -- Release team member -- Freelance Consultant

Attachment: signature.asc
Description: PGP signature


Reply to: