lilo vs grub

[Alexandre Oliva]

On Oct 21, 2003, Pekka Savola <pekkas at netcore.fi> wrote:

> Uhh.. is this really a problem?  If you re-add a disk after a failure, I'd 
> suppose the user would be intelligent enough to change the boot sequence 
> in the bios to actually skip hda :-).

Which may then change the disk numbering, or may not, depending on the
BIOS.

> Around RHL 6.1 or so (2-3 years ago), I did a lot of experimenting with
> software-RAID1 on root partition, including plugging off hda

That's easy, because then the disk is renumbered for sure.  Assuming
your raid1 /boot gets the same number as the original, it will
probably work.  What if it doesn't?  Say, I have these 8 hard disks on
this box; 4 built-in IDE channels, 6 add-on IDE channels, almost all
of them taken.  As soon as I added the PCI IDE controllers, the disk
numbering presented by the BIOS for the disks in the second built-in
IDE controller lost touch with reality, and the ability to tell the
BIOS to boot off this second IDE controller became mostly useless.
Luckily, I can still select which disk in the primary controller to
boot from, but I used to have /boot in disks in separate controllers,
just because I could, and I had to rearrange that because it no longer
works.  Fun, eh?

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer