RAID-1 setup doesn't boot on second Hard Drive

petrasl · 11-23-2008, 03:22 PM

Hello,

I just finished a Slackware-12.1 install with RAID-1. I used the
slackware-12.1/README_RAID.TXT how-to. The system just work.

I wanted to simulate a disk failure, so I halted the machine,
unplugged the first disk (/dev/sda) and re-booted.. And nothing: no lilo.

The 2 disk are SCSI and the controller is an Adaptec AIC-7899.

The /etc/lilo.conf is:

Code:

boot = /dev/md0
raid-extra-boot = mbr-only

[...]

image = /boot/vmlinuz-generic-smp-2.6.24.5-smp
  initrd = /boot/initrd.gz
  root = /dev/md0
  label = Linux-generic
  read-only

When I run "lilo"

Code:

Added Linux-generic *
The Master boot record of  /dev/sda  has been updated.
Warning: /dev/sdb is not on the first disk
The Master boot record of  /dev/sdb  has been updated.

thanks for help
petrasl

dkm999 · 11-23-2008, 10:32 PM

I am no expert on Slackware, but have you made sure that your BIOS knows that it should try to boot from the second hard disk if the first is not available? This sounds like what is going on here.

jayakrishnan · 12-22-2008, 11:16 AM

A question about raid here. If i setup linux using raid-1 with two harddisks, if the primary hard disk crashes, can i boot through the second hard disk?

mostlyharmless · 12-22-2008, 11:43 AM

Well, yes, that's usually one of the points of having RAID 1.

dkm999 · 12-22-2008, 12:48 PM

As with most things, it depends. The key here is to understand what actually happens during a boot sequence, and when the redundancy of your RAID array actually takes effect.

True hardware RAID works as soon as the disk controller is accessed, and so will allow boot-up when one drive is kaput. If the RAID setup is driven by software though, you will not be able to take advantage of automatic redundancy until that software is running. This means that you will have to make sure that several steps of the process will run from either disk:
1. Your BIOS needs to be informed that, if it cannot reach the first hard disk, it should try the second one.
2. Both disks in the RAID array should have a boot sector written on them.
3. This boot sector is very small; it generally needs to load a more capable block of software to complete the boot-up. GRUB and LILO are two examples of this. If your array is actually RAID-1, there will be a copy of the necessary stuff on each drive. But the boot sector on each drive needs to point at the right place on that drive for the boot process to be effective.

There are lots of posts on the web dealing with the grubby details of how to achieve these results, depending on your particular distribution and choices about how the RAID array is set up.