Red Hat Bugzilla – Bug 79896
grubby --bootloader-probe fails to find lilo with software raid
Last modified: 2007-04-18 12:49:07 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.2.1) Gecko/20021130
Description of problem:
Versions 21.4.3 and up of lilo have support for automatically working out where
to install the bootloader in RAID 1 setups. This allows you to set boot=/dev/mdX
(where mdX is the /boot partition) and have lilo automatically install the
bootloader to the boot sectors of the drives making up the RAID array. See
When lilo.conf includes the line boot=/dev/mdX, 'grubby --bootloader-probe' will
fail to return 'lilo' as expected since the bootloader has been installed to
/dev/hdX instead of /dev/mdX. If the lilo.conf file is changed so boot=/dev/hdX,
'grubby --bootloader-probe' works as expected (returning 'lilo').
This problem prevents kernel and other upgrades from reinstalling lilo when
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install Red Hat with a RAID 1 mirrored /boot partition.
2. Setup lilo using boot=/dev/mdX and install.
3. Run grubby --bootloader-probe which will not return anything.
4. Setup lilo using boot=/dev/hdX and reinstall.
5. Run grubby --bootloader-probe. This will now return 'lilo'.
Actual Results: Step 3:
Expected Results: Step 3:
I've just done a little investigation into the source for grubby and the problem
is that bootSect is read from the first 512 bytes of /dev/mdX, whereas lilo
actually installs itself to the boot sectors of the drives making up /dev/mdX
(e.g. /dev/hda and /dev/hdb). When boot is set to /dev/hda in lilo.conf, grubby
works correctly as bootSect is read from the first 512 bytes of /dev/hda.
do you have a correct /etc/raidtab sitting around?
Created attachment 89363 [details]
I've attached my current /etc/raidtab. With this RAID 1 setup 'grubby
this should be fixed in mkinitrd > 3.4.43 -- want me to send you an rpm to test?
http://people.redhat.com/~katzj/mkinitrd/ has the newer package.
Thanks. The new version returns the correct result on my RAID 1 setup using lilo.
Will this fix be made available as an errata package for Redhat 8.0?