If you are looking for the epic motorcycle journey blog that I've written, please see the Miles By Motorcycle site I put together. 
  • Solution: Can't Boot RHEL 3.x/4.x On RAID 1 Following A New Install
    04/15/2005 2:16PM
    This was one of the stranger OS install problems I've encountered in a while.

    Symptoms:

    After doing a fresh install of RedHat Enterprise 4.0 on a new Supermicro rackmount box, the system would come up to a text "GRUB" prompt and freeze. The behavior was reproducible with fresh installs of RHEL 3.x and Fedora 3.x as well.

    To make things even weirder, fresh installs set up to NOT boot on RAID1 would install and boot just fine.

    Nothing I could find via googling or the RH knowledgebase directly addressed the problem. I did work with RH tech support and they suggested reinstalling grub--which didn't work. Wiping out and recreating partions and swapping disk order didn't work either.

    Furthermore, following RAM and drive tests, and considering the fact that several other identically configured boxes were running RHEL 3.x/4.x without a problem, something didn't add up.

    Solutions:

    A second RH tech support suggestion,reinstalling Grub to both MBR on both primary boot drives in the RAID1, array did work.

    Subsequent fresh installs also worked after that. However, I will never deploy a server under non-emergency conditions without having as close to 100% confidence in it as possible. So rather than just deploying the box as is and calling it "fixed", I did one further test.

    I wiped the MBR and the partition table on all the drives in the box via Knoppix, and then did an additional fresh install of  RHEL4. This time, the box came up after installation without a hitch.

    Likely Explanation:

    Something in the MBR of one or both of the RAID1 boot drives was screwing up the boot process and was not getting eliminated via repartitioning and formatting.

    Reinstalling Grub to both drives in the RAID1 array and wiping the MBR and partition table on the afflicted drives both worked. The advantage to reinstalling grub on both members of the bootable RAID1 array means that if the primary drive dies you can still boot in degraded mode on the second member of the array.