Everything went fine, except that I messed up the lilo installation so
that when I rebooted my machine I booted the slackware partition
(/dev/hdb1) instead of the RedHat partition (/dev/hda1) that I had asked
for during the installation. ( I think I chose the wrong option during
the Fermi installation and wrote the new lilo to the boot record on
/dev/hda1 when I think I really wanted the MBR ... but I'm not sure. )
In any case, I thought I could fix the problem by running the slackware
version of lilo. I wrote this one to the MBR, and specified /dev/hda1 to
boot with label RedHat. This was the result :
-----------------------------------------------------------------------
Fermi Linux 6.11 (Strange)
Kernel Version 2.0.35
login :
-----------------------------------------------------------------------
So the wrong kernel got booted. This seems to me like a bug in the
slackware lilo that came with that kernel. It for some reason couldn't
find the 2.2.X kernel that came with Fermi RedHat and so it defaulted to
using whatever kernel it could find - namely the one it came with.
/dev/hda1 was still mounted as root of course, so I wound up with a broken
system.
Subsequently, I played around with moving vmlinuz here and about trying
to trick the boot loader, and ended up with a 'lost kernel' and an
unbootable system. I suppose that the boot loader needs to know the
beginning disk sector of the kernel and not it's pathname ?
I'm not sure what the lesson is here - except perhaps to never run an
alien version of lilo. And leave /vmlinuz alone. But I defer to the
experts to dispell any misconceptions I am spreading here.
Thank you very much,
Greg Graham
******************************************************************************
Greg Graham
ggraham@fnal.gov
University of Maryland
DZero Collaboration
630-840-2321
******************************************************************************