Bugzilla – Bug 116800
Installing grub rpm kills lilo
Last modified: 2007-09-27 15:50:32 UTC
When I installed grub.rpm for fun, it killed the MBR and put GRUB into it. That's definitely not what I wanted.
AFAIS the spec file does nothing with the MBR.
"Oh really? Impressive." :) Check this: 19:19 shanghai:~ # rpm -Uhv grub-0.96-5.i586.rpm Preparing... ########################################### [100%] 1:grub ########################################### [100%] 0+1 records in 0+1 records out 100948 bytes (101 kB) copied, 0.000838 seconds, 120 MB/s ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ (To answer: probably a backup of the old MBR, because grub is writing itself in there now.) And `rpm -qp --scripts grub-0.96-5.i586.rpm` also returns a lot, esp. In %postinstall: /usr/sbin/grub --batch < /etc/grub.conf cat /etc/grub.conf: 19:22 shanghai:~ # cat /etc/grub.conf root (hd0,0) install --stage2=/boot/grub/stage2 /boot/grub/stage1 d (hd0) /boot/grub/stage2 0x8000 (hd0,0)/boot/grub/menu.lst quit Q.E.D.
Do not draw premature conclusions. Hint: rpm -q --scripts ... Maybe you're actually using blocks belonging to deleted files? Will investigate next week when I'm back from brainshare.
Indeed. If there's a grub configuration, and grub is updated, it will install itself using the former. This is the easiest and most robust way to determine whether the user has opted for grub or lilo; it's not a bug, it's a feature. If you don't like it, move grub.conf aside (where did it come from, anyways?).
It came from the SUSE default install, and was leftover by the time I did 'rpm -e grub; rpm -Uhv lilo.rpm'. So in fact, I did not update grub (in which case this would have been ok), but a _new install_. AFAI know, rpm is intelligent enough to distinguish a %postinstall and %postupdate (whatever it's called), so I suggest placing it in the correct section.