:~>One of the updates was the 2.2.13-22mdk kernel. Now my machine won't
:~>boot properly. I've amended /etc/lilo.conf to point at the new kernel
:~>image, and run /sbin/lilo to update everything. When I re-boot, the
:~>machine gets as far as the "finding module dependencies" bit, then
Have you installed the new "kernel-modules" package?
:~>hangs. I can CTRL-C to continue booting up, but then I have no
:~>network card, no sound and no access to any of my non-Linux
:~>filesystems.
:~>
:~>I promise that if someone can help me get this fixed, I won't
:~>recklessly apply OS updates ever again without checking the
:~>documentation and/or making backups first!!!
Updating the kernel is a bit different from updating most of the
other other packages.
Basically, if you screw things with "normal" programs, you can usually
stil boot (at least in single user mode) and repair it. On the other hand,
screwed kernel instalation often leaves you with a system which cannot
boot at all which is rather annoying (where have i put the "rescue disk?")
.
Next time you decide to make the kernel update, try something like this
(*):
1) Install the new kernel, but leave your old kernel where it is.
Same with kernel modules. I THINK you can do this by applying "rpm -i"
instead of "rpm -U". It certainly works if you compile the kernel yourself
2) Make a new initrd. Something like:
mkinitrd /boot/initrd-2.2.13-22mdk.img 2.2.13-22mdk
3) add a new entry to "/etc/lilo.conf". Something like:
image=/boot/vmlinuz-2.2.13-22mdk
label=new
root=/dev/hda5
initrd=/boot/initrd-2.2.13-22mdk.img
read-only
4) run lilo.
On reboot, choose "new" - if it works properly, you can rename the "new"
to "linux" and the old "linux" entry to "new", run "lilo" and reboot
again. In case anything goes wrong, you can still use the old one...
Hope this helps
Denis
(*) THERE IS MORE THAN ONE WAY TO DO IT
