How do you recover when grub boots to 'grub>'?

Hi,

I actually know most of the answer to my question, but I am stuck on one point.

If grub.conf is corrupt a server will boot to 'grub>'. To boot the server you need to build the root, kernal and initrd lines which would normally appear in grub.conf. I know how to figure out everything excpet the dev device for the boot partition which is needed for the 'root=' parameter on the kernel directive.

I can find and cat /etc/fstab but I'm on Centos6 and it only lists the UUIDs of the partitions.

I could try each dev device one by one until I eventually select the right one but I was hoping there is a more sensible approach. Sadly I've search online but not found a solution to this yet.

Grateful for any advice on this.

Gareth

root=UUID=<UUID number>

Thanks. I forgot you could do that. Is there no way of translating it though as I seem to mistype the UUID frequently when performing these recovery tests.

I don't think so, but not for sure, someone on here with more brains(of course it don't take a whole lot to pass me up in brain power LOL) then me might know that answer.

Live cd's help find out stuff.

Any chance you figured this out? I have a similiar problem if you could help me out. Thanks.

http://www.linuxquestions.org/questi...issues-943498/

thewird

Hi,

I concluded there's no way to translate a UUID to device in grub. Whilst tune2fs and blkid offer this, it's obviously not available whilst you're in grub. Therefore the realistic options are:

1. Use the UUID for / in 'cat (hd0,X)/etc/fstab'
2. Trial and error until you hit the correct disk
3. Keep a record of important configuration info of your servers off the system which incudes the UUID maping, which can be used in such emergencies.

From a support perspective I think option 3 would be good practice but may rarely get done.

Thanks again
Gareth

grub2 might be your saviour.