Problem booting OS on IDE PCI adaptor card with Grub in MBR of hda
Mikkel L. Ellertson
mikkel at infinity-ltd.com
Mon Jul 17 01:45:10 UTC 2006
>> Hi Tim. My confusion was in not realising that Grub counted the harddrives
>> sequentially, ignoring how they were identified in /dev. This wasn't helped
>> by the fact that /dev/hda, and /dev/hdb, the first 2 harddrives, were listed
>> by Grub as (hd0), and (hd1), which at the time seemed to confirm that Grub
>> followed the listings in /dev. Logically then /dev/hde is (hd4) right? No.
>> Very wrong, and resulted in the long running argument with Grub, with it
>> constantly saying "no such disc" when I tried install Grub in (hd4).
>
> Yes, it is irritating, though explained in the documentation. It's a
> right pain on systems with removable drives. You can't take a drive out
> and leave it out, unless all the GRUB configurations come before it.
>
> i.e. On a five drive system, you couldn't take out the second drive, and
> have the following three drives treated the same as before. You can
> only take out drives from the last drive, backwards.
>
Lilo is supposed to handle this better now if I am reading the man
page correctly. Here is the section from "man lilo.conf".
NOTE: Use of the 'bios=' option is largely obsolete beginning
with LILO version 22.5, as the boot loader now identifies disks
by 32-bit Volume-ID, and defers BIOS device code determination
until boot time.
But I have not played with it to see what happens if you chainload
to another copy of lilo on the MBR of another drive.
Mikkel
--
Do not meddle in the affairs of dragons,
for thou art crunchy and taste good with Ketchup!
More information about the users
mailing list