Re: [tlug] Kernel panic

["SL Baur" <steve@example.com>]

On 4/23/08, Stephen J. Turnbull <stephen@example.com> wrote:
> Brian Chandler writes:
>
>   > Plenty (gigabytes) of space, so I deleted the empty file.

>  Point the Third: Unless *you* truncated that initrd, something in the
>  install or upgrade process did it (nothing else is at all likely to
>  touch an initrd).  "Just reinstalling the bits" would truncate it
>  again....

That's a disturbing point.  I'd want to know why he got a zero
length file.  The only simple explanation is out of disk, but since
he's got plenty of disk, that can't be it.  The version of mkinitrd
that I have refuses to create a file when an error occurs *and*
the initrd *is* generated at installation time (ie. there's no initrd
in the kernel RPM).

Actually, the initrd has to be generated at installation time because
that's where the few vital kernel modules get loaded and which ones
those are are highly system dependent.

Brian, when you were looking for mkinitrd and yaird were you looking
in the rescue environment or on your mounted system disk?  This
isn't an obvious looking case of filesystem meltdown so you ought
to be able to chroot into your system disk and run things that way.
If for example, your system disk is mounted on /mnt/root do:

cd /mnt/root
chroot /mnt/root

(Substitute the actual mount point for /mnt/root)

Then look for /sbin/mkinitrd, etc.

If they haven't put chroot on your rescue disk, that would be
criminal.

-sb