I initially installed this server (SME 7.3) with 3 disk in a RAID5 config (using the nospare option) I recently got a real disk failure and a false disk failure....at the same time.
This is a production server, so i panicked (and so did the kernel).
At boot I was now getting an error stating that md2 couldn't start with 1 of 3 drives, which makes sense.
I used a Knoppix cd to rebuild the array, swapped out both reportedly bad drives just to be safe and continued on. All the data is there. Happy day. But here is where the trouble comes in. I must have messed up somewhere down the line becuase the UUID for md2 got changed. Now, trying to boot the SME system, I get
md: md2 stopped
mdadm: no devices found for /dev/md2
and then a kernel panic.
So, I put in my trusty SME 7.3 disk and go into rescue mode. It has no problem automatically detecting the array, and i can get at the data no problem. So why wont it auto configure it from the boot off md1?
I've tried specifying the new uuid in the mdadm.conf, changing the kernel parameter "md=" to force it. I am running out of ideas.
And now the question.
Is there anyway to tell the kernel the new uuid? Will it help if i regenerate the initrd and boot img for the kernel? Is that where it is stored? And if so...how?
I can provide more detail as to the steps where i messed up, if needed.
I REALLY appreciate any help on this matter, thanks in advance.
- Dan
[SOLVED]
I also ran into a grub error, sinc eI had swithced out /dev/sda...my boot device...complete solution
1. from the install disk I typed "sme rescue" to go into rescue mode.
2. Allowed the loader to detect and mount my raid array and lvm volumes
3. chroot into the mounted existing installation per instructions of the SME boot cd
3. changed /etc/mtab to point /boot to /dev/sda1 from /dev/md1
4. ran grub-install /dev/sda
5. ran mkinitrd /boot/initrd-2.6.9-67.0.7.ELsmp2.img 2.6.9-67.0.7.ELsmp
6. then i rebooted gracefully
7. at reboot i went into the grub menu and manually changed the initrd image to my new one....
BOOTED!