Re: Broken initrd image

2007/8/24, Émeric Maschino <emeric.maschino@gmail.com>:

Hi Mike,

Thanks for your answer.

Partition usage on /boot is only 25%. Enough for hosting at least 3 different initrd images.

BTW, I just remember that I had briefly installed gcc-2.96-base with libstdc++5 in order to test an old software. As stated in my previous email, I've nonetheless reinstalled all the current gcc-4.1-base and related glibc and libstdc++ since then.

Émeric

2007/8/23, Mike Bird < mgb-debian@yosemite.net>:
On Wednesday 22 August 2007 15:23, Émeric Maschino wrote:
> I'm running up-to-date Debian Lenny on my hp workstation zx6000. Today,
> initramfs-tools was updated and a new initrd image was created by the
> update-initramfs script. Unfortunately, the generated initrd image
> segfaults and my system was unbootable. Using the backup copy of my
> previous initrd image, everything went fine.
>
> I initially thought that the new initramfs-tools 0.90 was broken. But
> reverting to initramfs-tools 0.89 and generating again the initrd image
> didn't solve the problem. So, something has changed on my system that makes
> update-initramfs (being 0.89 or 0.90) produces unbootable initrd image. The
> verbose option doesn't show any problem.
>
> Roughly one year ago, I experienced the same problem. But it was due to
> incorrect ldd binary. This doesn't seem to be the case this time, as the
> output of ldd /bin/dash is fine. By security, I've reinstalled all the gcc,
> cpp, glibc and libstdc++ related packages. Still unbootable initrd image
> though.
>
> If it can help diagnose the problem, all the packages on my system are .deb
> packages coming from the Debian Lenny repository, except for the Intel
> development toolchain (C compiler, debugger, math kernel library, Intel
> performance primitive library).

How are you doing for space in /boot? I've seen several systems hosed
because they ran out of space for the initrd after something changed in
Debian to keep additional backup copies in /boot.

--Mike Bird