Your message dated Thu, 13 Aug 2009 14:12:38 +0200
with message-id <1250165558.3517.5.ca...@fz.local>
and subject line Re: Bug#541286: Update-grub guesses wrong root
has caused the Debian Bug report #541286,
regarding Update-grub guesses wrong root
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact ow...@bugs.debian.org
immediately.)
--
541286: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=541286
Debian Bug Tracking System
Contact ow...@bugs.debian.org with problems
--- Begin Message ---
Package: grub-pc
Version: 1.96+20090725-1
Severity: critical
Justification: system unbootable after update
Hi,
after updating grub-pc, the system was unbootable. The problem is, that
update-grub is run in postinst and it guesses the wrong boot partition.
I have a mixed IDE/SATA system and the disks are ordered differently by
the BIOS and Linux.
For this reason I have set the label SYSBOOT on the boot partition and
added
set root=SYSBOOT
to /boot/grub/grub.cfg. This works like a charm, however:
- update-grub is called in postinst and overwrites grub.cfg (naughty!)
- I haven't found a way to configure the boot partition for update-grub
to use.
I think the only real solution is for update-grub to stop thinking it'd
be smarter than it actually is. A lot of automagic is done which broke
my boot a view times already. There is no straightforward way to tell
grub to do as I say and make it obey its master's wishes. This `smart'
system made me do too much working around and circumventing, already.
Combined with the fact that it is also very inflexible (e.g. only one
kernel cmdline for all entries), it is my opinion that it should be
replaced by something very different, which does less guessing and
gives more power to the user.
If you don't agree, at least make it configurable so I can prevent
update-grub from ever running. Please let me configure my system as I
like.
Cheers,
harry
signature.asc
Description: PGP signature
--- End Message ---
--- Begin Message ---
Am Donnerstag, den 13.08.2009, 11:46 +0200 schrieb Harald Braumann:
> On Thu, 13 Aug 2009 07:00:48 +0200
> Felix Zielcke <fziel...@z-51.de> wrote:
>
> > Am Donnerstag, den 13.08.2009, 02:57 +0200 schrieb Harald Braumann:
> > > Hi,
> > >
> > > after updating grub-pc, the system was unbootable. The problem is,
> > > that update-grub is run in postinst and it guesses the wrong boot
> > > partition. I have a mixed IDE/SATA system and the disks are ordered
> > > differently by the BIOS and Linux.
> >
> >
> > That's why you sometimes have to fix /boot/grub/device.map yourself.
> > It gets only recreated if you run grub-install with --recheck or
> > directly grub-mkdevicemap.
> > But anyway we use UUIDs in the default grub.cfg so it shouldn't matter
> > how your disks are ordered in device.map
> > The search --set --fs-uuid line sets the root variable to the device
> > with that UUID.
> >
>
> Something else must have been rotten, then. When I first booted after
> the update, I was dropped in grub rescue immediately. The error message
> was something like "symbol not found bz_<something>". None of the
> commands, like `help', `search', etc. worked. Error was always
> <command> not found.
>
> I booted from USB and changed grub.cfg. I replaced "set root=(hd...)"
> with "set root=SYSBOOT". Rebooted and it worked. So I was sure that
> this was the cause.
Weird. Maybe the embed core.img was out of sync with the rest
of /boot/grub, but then simple chaning root shouldn't fix this.
About missing symbol grub_zalloc there's a bug in Launchpad:
https://bugs.launchpad.net/ubuntu/+source/grub2/+bug/408699
> However, I can not reproduce the error now. I tried to change grub.cfg
> back to how it was before, but that works. I called update-grub and
> that works, too.
>
> I guess we have to close that bug as unreproducible.
>
Ok closing.
--- End Message ---
