From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)
The disk Redhat 7.0 was being installed on was given the 'hdb' label
instead of 'hda'. A CD-rom was also attached with the same cable which is
why it was not chosen as the first disk. At the end of the installation
procedure, Redhat ran lilo and then rebooted. The only problem was that
lilo failed because it was being installed on the second disk. Instead of
Redhat noticing that lilo failed, and at least giving the user the option
to edit lilo.conf, it blindly rebooted. Of course, upon rebooting, there
was no boot loader installed, so Redhat did not boot.
It took me a long time to finally get hdb booted as root, edit lilo.conf,
and then run lilo myself. If I were given the option to edit lilo.conf
just before Redhat rebooted, it would have saved hours of work.
Therefore, it is my opinion that if lilo fails, the Redhad installer
should consider this a SERIOUS problem and not reboot. In the best case,
if the installer notices that lilo failed, it can give some sort of user
friendly menu of possible problems it can try such as 1) installing on the
second disk 2) installing past the 1024 cylinder boundary. At the very
least, the installer should allow the user to modify lilo.conf to get lilo
working before rebooting. If it blindly reboots the installer should know
that Redhat will not boot and the user will be stuck. There is no point
to reboot if the boot loader is not installed.
Steps to Reproduce:
1. Install Redhat on the second disk (hdb) OR install Redhat past the 1024
Actual Results: Upon reboot, Redhat Linux didn't start.
Expected Results: Upon reboot, Redhat Linux should have started.
We have added a screen when lilo fails, indicating the user should make a boot
floppy (the next screen in the installer) so they can get into the installed
system and fix the lilo.conf.