up vote 0 down vote favorite
share [g+] share [fb]


I have a debian with a raid1 on it. Both HDDs have bootable flags and grub setup in MBR. If I only start up with sda, linux boots. If I only start up with sdb, grub shows up and says 

Decompressing Linux...
crc error
-- System halted

I have reinstalled grub a few times now but still nothing.
It goes like this:
/dev/md0 contains /dev/sda1 /dev/sdb1
find /boot/grub/stage1 shows (hd0,0) (hd0,1)
The line that boots Linux shows:
root (hd0,0)
kernel /vmlinuz-2.6.26-1-686
initrd /initrd.img-2.6.26-1-686

This kind of renders my "redundant" array useless.
Any clues?

update: just to mention, these are 2 different HDDs, sda is 320gb and sdb is 400gb. Both are WD and both have exactly the same partitions, cloned using sfdisk.

link|improve this question
64% accept rate
Is this a new installation or has it been working for some time ? – Iain♦ Jun 13 '10 at 10:41
I added the 2nd hard drive last week, but didn't had to test it until yesterday. – w00t Jun 13 '10 at 17:12
feedback

3 Answers

up vote 1 down vote accepted

I finally managed to repair it. It was a HDD related problem and I had to go into BIOS, at the HDD's options and disable Block Mode.
Then I just proceeded with the normal installation: re-add the partitions to the array, setup grub.
It now works booting from this drive.

link|improve this answer
feedback
up vote 0 down vote 
root (hd0,0) 
kernel /vmlinuz-2.6.26-1-686 
initrd /initrd.img-2.6.26-1-686

If this is from your grub.conf on sdb, it's missing some important information, such as what to boot. The kernel line should look something more like this:

kernel /vmlinuz-2.6.26-1-686 ro root=/dev/sda5

with root= pointing to the block device that contains the / filesystem.

In addition, if you're booting off the secondary drive, you'll want to specify that grub actually use it -- specifically,

root (hd1,0)
link|improve this answer
Well, considering the scenario where sda fails, I remove it and sdb now becomes sda, so hd1 is hd0, I don't need to modify that. I also don't need the root= because it already knows where the root is -> root (hd0,0). The problem is that it finds vmlinuz and initrd, starts decompressing but gives that crc error. – w00t Jun 13 '10 at 10:28
feedback
up vote 0 down vote

If you built the raid correctly and it fully synced then chances are high that you have some bad sectors on sdb.

What does the output of 

cat /proc/mdstat

say about the device ?

link|improve this answer
It used to be in sync but now I have kicked the drive out of the array. I'm trying to use separate /boot partitions for each drive and keep raid just for / – w00t Jun 15 '10 at 6:51
I just installed a fresh linux on the ex-sdb and even with this one grub gives errors - error 18, which I googled and somebody said: "18 is a hard-disk read on the wrong cylinder or such". – w00t Jun 16 '10 at 21:22
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.