I had a close call with my promise-sata RAID-1 drives today. The machine had been stable for several weeks now configured with software RAID-1 across two identical SATA drives. Both drives have RAID-1 partitions for /var, /boot, /, swap and /home. Earlier this week I followed the instructions at... http://www.dirigo.net/tuxTips/avoidingProblems/GrubMdMbr.php to allow the second SATA drive to be bootable if the first drive failed. I followed the instructions... device (hd0) /dev/sdb root (hd0,0) setup (hd0) quit within grub. The machine seemed fine for several reboot. Today however I found it frozen and had to do a hard reboot. The machine refused to load the boot loader claiming stage2 was missing. I was able to get the 'linux rescue' mode to find the RAID-1 partitions and mount them as /mnt/sysimage. However I found that after a 'chroot /mnt/sysimage' that executing 'grub-install /dev/sda' couldn't reinstall the boot loader on the MBR because grub claimed that the BIOS didn't have md devices. At this point I tried another reboot and fortunately the boot loader finally seemed to work this time and I was able to boot into the kernel. Executing 'cat /proc/mdstat' shows that md1 for the / partition is resyncing. My questions are... 1) are those instructions for making the second drive bootable incorrect? 2) how exactly is one supposed to reinstall the MBR for a RAID-1 when in linux rescue mode? I wonder if I should have done... device (hd0) /dev/sda root (hd0,0) setup (hd0) quit from within grub rather than just 'grub-install /dev/sda'? Jack
