Reading this month's XFS list archive shows a couple posts regarding GRUB
problems in the XFS/Redhat installer.
FWIW, we used the XFS 1.0 / Redhat 7.2 installer on probably a dozen
machines of varying spec (laptops up to Compaq servers with hardware RAID),
all with GRUB, and it worked fine.
Yesterday we did our first XFS 1.1 / Redhat 7.3 install, and GRUB failed
miserably - loads stage1, then dumps to a grub> prompt. A coworker did
another install today with the same result. In both cases I was able to
fix it by booting the CD in rescue mode, and manually mounting the
partition containing /sbin (in this case, /dev/hda2). /boot is hda1.
Then, I ran grub from there (/mnt/foo/sbin/grub) and did:
grub> root (hd0,0) (sets GRUBs root to /boot, /dev/hda1)
grub> setup (hd0) (reinstall GRUB into MBR)
grub> exit
# cd / ; umount /mnt/foo
(ctrl+alt+delete, eject your CD)
Note that you don't want to do "setup (hd0,0)" or it will try to write the
bootblocks over top of /boot (you can gues how I found that out...)
These installs were on fairly modest Compaq desktops with IDE drives... the
maillists archive leads me to believe that this problem and the 1.0/7.2
problems are somewhat machine dependent, and that the sync; sleep 30; sync
hack isn't the right way to address it. (Feels like the kernel timer that
is used by the VM and the sleep() timer are not at all aligned...)
Sometime next week I'll have time to do a fresh install on a Compaq DL360
with hardware RAID, we'll see how that goes.
--
Jameel Akari
UNIX Admin
First Albany Corp
|