problems with software RAID on 2.6.3 and 2.6.4.
I recently installed a system using RAID 1 via the SuSE install procedure. On 2.4.21-193, everything's fine, but after compiling 2.6.3 (and then 2.6.4 to see if that'd fix it), it sees the md devices and attempts to activate them, but fails to mount the root FS. I was wondering what's changed in 2.6 to cause this, and if there's a workaround? (Can't give much more info cos it all scrolls off the screen too fast and once the kernel's paniced with "Failed to mount the root fs", I can't scroll back to see if there's a specific complaint.) Thanks.
Hi! Maybe naming of devices? 2.4 sees SATA devices as /dev/sda, /dev/sdb ... 2.6 sees SATA devices as /dev/hda /dev/hdb So init trie to boot up let say from /dev/sda1 but since there is new kernes, reports first disk as /dev/hda1 So it fails to mount root. Try to add root=/dev/sda1 or other boot partitions that you use and then try to change /etc/fstab file. Hope this helps. MIaln Andrew Halliwell wrote:
I recently installed a system using RAID 1 via the SuSE install procedure. On 2.4.21-193, everything's fine, but after compiling 2.6.3 (and then 2.6.4 to see if that'd fix it), it sees the md devices and attempts to activate them, but fails to mount the root FS.
I was wondering what's changed in 2.6 to cause this, and if there's a workaround?
(Can't give much more info cos it all scrolls off the screen too fast and once the kernel's paniced with "Failed to mount the root fs", I can't scroll back to see if there's a specific complaint.)
Thanks.
And verily, didst Milan Gabor announce to the hordes:
Hi!
Maybe naming of devices? 2.4 sees SATA devices as /dev/sda, /dev/sdb ... 2.6 sees SATA devices as /dev/hda /dev/hdb So init trie to boot up let say from /dev/sda1 but since there is new kernes, reports first disk as /dev/hda1
Ah, sorry... This is standard ATA, not SATA. (software raid on /dev/hda2, /dev/hdb2 for the / partition (all others apart from /boot (/dev/hda1) are too, of course) That's one of the reason's why it's so puzzling.
On Sat, 13 Mar 2004, Andrew Halliwell wrote:
Ah, sorry...
This is standard ATA, not SATA. (software raid on /dev/hda2, /dev/hdb2 for the / partition (all others apart from /boot (/dev/hda1) are too, of course)
That's one of the reason's why it's so puzzling.
I was going to say I had exactly the opposite problems... Can you view the /var/log/messages and /var/log/boot.log from the 2.6.x kernels? Try the boot, then use the rescue disk to read those files, or move them so they won't be clobbered by your 2.4 kernel. You might have hints in there. Patience, persistence, truth, Dr. mike
And verily, didst Mike Rosing announce to the hordes:
This is standard ATA, not SATA. (software raid on /dev/hda2, /dev/hdb2 for the / partition (all others apart from /boot (/dev/hda1) are too, of course)
That's one of the reason's why it's so puzzling.
I was going to say I had exactly the opposite problems...
Can you view the /var/log/messages and /var/log/boot.log from the 2.6.x kernels? Try the boot, then use the rescue disk to read those files, or move them so they won't be clobbered by your 2.4 kernel. You might have hints in there.
Thanks for the idea, but... Alas, as I suspected, no logs. As the root filesystem isn't getting mounted, it can't mount /var for log writing either.
On Sat, 13 Mar 2004, Andrew Halliwell wrote:
Thanks for the idea, but... Alas, as I suspected, no logs. As the root filesystem isn't getting mounted, it can't mount /var for log writing either.
What modules do you need for raid to work? I have avoided learning that on purpose :-) My bet is that you need to tell the initrd to create a different ram module set than you have now. It's not going to be easy to figure it out. Can you use the interactive boot, or don't you even get that far? Patience, persistence, truth, Dr. mike
participants (3)
-
Andrew Halliwell
-
Mike Rosing
-
Milan Gabor