Re:error on LILO after RH 7.1 up2date

> > I just ran up2date on my 7.1 box and elected to only update the kernel and
> > xinet.  The Kernel went from 2.4.18.X to 2.4.20.X.  I received an error at
> > the finish,  "LILO test failed"  this is not an exact quote.
> >
> > This box has been up and running for around 3 years and scheduled to be
> > superceded this summer, I have done a lot of patching on it over the years
> > and need to keep it up for another 4 to 8 weeks.
> >
> > Recently because I was running out of space in the partition I went into
the
> > /boot section and deleted a lot of really old versions of the kernel.  I
> > guess this has screwed some index up.  They are still listed in the boot
> > menu.
> >
> > Any ideas on how I clean this up?  I am terrified to reboot it. (Thank
> > heaven for my UPS)
> >
>     Edit /etc/lilo.conf so that it doesn't have sections that refer to
> files you've deleted.  Then run lilo.  "lilo -tv" will test whether
> lilo.conf seems ok without actually installing the boot loader.
>     If you have a boot floppy you have an alternative way to boot if your
> normal way fails.  If you don't have a boot floppy, type "man mkbootdisk".
>

In future, instead of deleting files manually like that, do the following:

rpm -qa | grep kernel

>From this, select all the packages associated with the old version(s) you want
to get rid of.  For EACH of these, remove the old version from your system
with:

rpm -e package_name

Then edit /etc/lilo.conf, as described above.

Don't forget to add a new /etc/lilo.conf section that refers to the latest
version you have installed!

Run "lilo -v" before rebooting.

Personally, I always keep one old version available, in case the new one
doesn't boot successfully.  So far, I have not needed it.  But the first time
I don't do this...  ;)

Regards,
Jim