Sorry, my last post was maybe a bit ambiguous. When I wrote "/dev/sda?" I didn't mean to replace the question mark with the partition number, I was just asking whether it is /dev/sda or /dev/sdb or ...

So try again with

grub-install /dev/sdb

Usually you install the boot loader in the MBR of a hard disk, and not in the boot record of a partition. The latter you only do when you chainload from another boot manager.

Yeah, I put sdb instead of sdb3
Apparently it worked.

But the system don't want to boot.
It says "File not found" when I choose from grub menu "CentOS 2.6.32-431.el6.x86_64"

Attached Thumbnails

 

I'm not a grub expert, so maybe someone else can step in. But I think we're on the right track. BTW are we talking about grub 1 or 2? Since your grub config was working before I wouldn't think that you need to recreate the grub.cfg. When in the grub menu, can you press "e" and write here what it says?

Yes... definitely we are on the right track!
I put a picture with what e key returns on grub menu.

hd1,2 - I changed that ... maybe I will gues the right numbers, sincer I don't know if it's 0 or 1, and what partition number is right.

Attached Thumbnails

 

Hmm... we have progress!!

I changed in grub.conf... from hd1,2 in hd0,2... and now the os want's to boot. The centos logo appears.
After a while it says:

Kernel panic - not syncing. Atempt to kill init!
PID1 init not tained 2.6.32-573.el6.i686 #1
Call trace:
...... switching back to text console.



NOW what? So close to success!

You know that you can change the grub boot parameter from within grub? You don't need to change your grub.cfg every time. Just edit the line with 'e' and when you finished editing press 'b' to boot. Much quicker than booting rescue, mounting and editing grub.cfg.

I found this thread where someone had the same problem. Some said that there is a problem with certain kernel versions, someone solved it by replacing 'root=UUID=...' with 'root=/dev/md0'. I too suspect that the UUID is the one from the boot partition of the failed drive. You can check the UUID by booting into the rescue system and issue the 'blkid' command. This should give you the UUIDs of all partitons. So use the correct UUID or try 'root=/dev/md124' or 'root='/dev/sdb3'. Keep in mind that sdb might not be sdb when you don't boot from the rescue media. You might also need to rebuild the initramfs when you are in the chroot after changing the grub.cfg.

The way it has to be done is to either run grub2-mkconfig /dev/xxx -o /boot/grub2/grub.cfg (if on a non-efi system. If there was never a grub segment in the mbr of that disk, you may have to run

grub2-install /dev/xxx

Please read info grub2.

Nope. CentOS 6 doesn't use grub2.

From a liveCD go here and download the script. Run it as the readme says. Post the RESULTS.txt here so we can see what the real situation is in your environment rather than guessing.

When I run the script ./bootinfoscript it says:

"./bootinfoscript: /bin/bash^M: bad interpreter: No such file or directory."