Migrating a Linux boot partition to RAID-1

• Shut down your computer
• Move your current root drive (which we will refer to as the old disk) from master on the primary IDE bus (/dev/hda) to master to the secondary IDE bus (/dev/hdc).
• Install the new disk that you will be using as the first drive of the array as the master on the primary IDE bus (/dev/hda).
• Boot the system. As the computer starts up, hold down the Control key until you get the lilo boot prompt. At the prompt, type linux boot=/dev/hdc root=/dev/hdc1 where /dev/hdc reflects the location of the old drive. Once the computer has booted, log in as root.
  You may get mount errors during boot if you have additional partitions listed in /etc/fstab that have moved to different locations or that have been removed. If you do, edit /etc/fstab to reflect their new locations or simply comment them out and then reboot your machine (If you do so, be sure to specify the boot=/dev/hdc and root=/dev/hdc1 options while booting).
• Create an /etc/raidtab with the array intitially in degraded mode:

  raiddev                /dev/md0
  raid-level             1
  nr-raid-disks          2
  nr-spare-disks         0
  persistent-superblock  1
  device                 /dev/hda1
  raid-disk              0
  device                 /dev/hdc1
  failed-disk            1

  /dev/hda1 and /dev/hdc1 should reflect the physical locations of your new and old drives, respectively.
  The persistent-superblock tells the kernel to write meta-information about the array to each partition that is part of the array. This provides protection against drive mix-ups and also allows the array to boot even if one of the member drives has failed.

  By marking /dev/hdc1 as failed, we are telling the kernel not to use it as part of the array. This allows us to continue using our old, non-RAID boot disk while we set up the array.
• Create a copy of the existing lilo.conf by running cp /etc/lilo.conf /etc/lilo.conf.raid Modify lilo.conf.raid to allow booting off of the array. The differences from a non-RAID lilo.conf are:

  # Specifies the boot device.  This is where lilo installs its boot
  # block.  It can be either a partition, or the raw device, in which
  # case it installs in the MBR, and will overwrite the current MBR.
  boot=/dev/md0
  
  
  # Specifies the location(s) within the array to write boot blocks.
  # If you want to have your array bootable off of any disks besides
  # the primary then you should specify the additional disks here. 
  # Note that not all BIOSs support this.
  raid-extra-boot="/dev/hda,/dev/hdc"
  
  # Specifies the device that should be mounted as root. (`/')
  #
  root=/dev/md0
  
  
  # This is the section that allows booting off of the RAID array.
  # The append line tells lilo about the geometry of the array.  
  # Ir describes the array as being named md0 and constisting of 
  # two partitions, /dev/hda1 and /dev/hdc1
  image=/vmlinuz
          label=Linux
          read-only
          restricted
          append="md=0,/dev/hda1,/dev/hdc1"
  
  
  # For convenience while setting up the array, you may wish to add
  # the old boot partition as show below.   Be sure to remove this
  # once the array has been set up since accessing the array 
  # partitions directly will cause serious damage to the array. 
  image=/vmlinuz
          label=LinuxNoRaid
          read-only
          restricted
          append="root=/dev/hdc1 boot=/dev/hdc"
  

  Note that RAID device names rather than the physical device names are given for the boot and root options. In addition, both lilo and the kernel need to be told the geometry of the array. This is done with the raid-extra-boot option (which tells lilo where to write the bootsectors) and the append option (that tells the kernel about the array). Note that if you already have an append line specified, you should add md=0,/dev/hda1,/dev/hdc1 to the exisiting line.
• You may wish to backup the old drive's master boot record and partition table. You write these to a floppy disk by running dd if=/dev/hdc of=/dev/fd0 bs=512 count=1 If anything goes wrong, you can restore the old MBR and partition table by running dd if=/dev/fd0 of=/dev/hdc bs=512 count=1
• Determine size of the old drive:
  • Run fdisk /dev/hdc
  • Type uEnter to change the units to sectors
  • Type pEnter to print the drive's partition table
  • write down the start and end sectors of the root partition (/dev/hdc1 in this example)
    For example, if the output of the p command was:

    Disk /dev/hdc: 120.0 GB, 120034123776 bytes
    255 heads, 63 sectors/track, 14593 cylinders, total 234441648 sectors
    Units = sectors of 1 * 512 = 512 bytes
    
       Device Boot      Start         End      Blocks   Id  System
    /dev/hdc1   *           1   234436544   117218272   83  Linux
    

    then you should write down 1 and 234436544.
  • Type qEnter to exit the program.
• Clear the partition table from the new drive by running dd if=/dev/zero of=/dev/hdc bs=512 count=1
• Partition the new drive:
  • Run fdisk /dev/hda
  • Type uEnter to change the units to sectors
  • Type cEnter to leave DOS compatibility mode
  • Type nEnter to create a new partition
  • Type pEnter to select a primary partition
  • Type 1Enter to select the first partition
  • Type 1Enter to set the first sector of the partition
  • Enter the start sector that you recorded above and then hit Enter.
  • Enter the end sector that you recorded above and then hit Enter.
    If you get an error saying Value out of range then your new drive is too small.
  • Type wEnter to write the partition table to disk and exit the program.
• Reboot the system. As the computer starts up, hold down the Control key until you get the lilo boot prompt. At the prompt, type linux boot=/dev/hdc root=/dev/hdc1 where /dev/hdc reflects the location of the old drive. Once the computer has booted, log in as root.
• Build array by running mkraid /dev/md0 You should see something similar to the following (the size and superblock information will vary):

  handling MD device /dev/md0
  analyzing super-block
  disk 0: /dev/hda1, 4210888kB, raid superblock at 4210816kB
  disk 1: /dev/hdc1, failed

  It is normal that disk 1 is shown as being failed since we have explicitly marked it as such. This is so that we can continue to use the /dev/hdc1 partition as the root partition while we finish setting up the array.
• Check that the array is set up correctly by running cat /proc/mdstat. You should see something similar to this (again the sizes will differ):

  Personalities : [raid1]
  read_ahead 1024 sectors
  md0 : active raid1 hda1[0]
        4210816 blocks [2/1] [U_]
  
  unused devices: <none>

  This shows that the array has been built but is degraded and that it currently is comprised of a single partition, /dev/hda1.
• Create a filesystem on the array by running mke2fs -j /dev/md0 This will create an ext3 (ext2 with journal) filesystem. If you want, you can use a different filesytems (ReiserFS, etc.) on the array Instead of ext3.
• Test that you can mount the array by running mkdir /raid and then mount /dev/md0 /raid The filesystem should mount cleanly. When you have verified that it is mountable, run umount /raid