4. Configuring the Production RAID system.

4.1 System specs.

Motherboard:    Iwill P55TU     dual ide + adaptec scsi
Processor:      Intel P200
Disks:          2 ea. Maxtor 7 gig eide

4.2 Partitioning the hard drives.

Since testing a large root mountable RAID array is difficult because of the re-boot problem, I re-partitioned my swap space to include a smaller RAID partition for testing purposes. You may find this helpful.

        <bf/DEVELOPMENT SYSTEM/

  /dev/hda1     dos             16meg
* /dev/hda2     extended        126m
  /dev/hda3     linux           126m    root partition during development
  /dev/hda4     linux           6+gig   raid1
* /dev/hda5     linux            26m    test raid1
* /dev/hda6     linux swap      100m

  /dev/hdc1     is simply an exact copy of hda1 so the
                partion can be made active if hda fails
* /dev/hdc2     extended        126m
  /dev/hdc3     linux           126m    /usr/src  during development
  /dev/hdc4     linux           6+gig   raid1 mirror
* /dev/hdc5     linux            26m    test raid1 mirror
* /dev/hdc6     linux swap      100m

        <bf/PRODUCTION SYSTEM/

/dev/hda1       dos             16meg
/dev/hda2       linux swap      126m
/dev/hda3       linux swap      126m
/dev/hda4       linux           6+gig   raid1

/dev/hdc1       is simply an exact copy of hda1
/dev/hdc2       linux swap      126m
/dev/hdc3       linux swap      126m
/dev/hdc4       linux           6+gig   raid1 mirror

The hdx3 partitions were switched to 'swap' after developing this utility. I could have done it on another machine, however, the libraries and kernels are all about a year or more out of date on my other linux boxes and I preferred to build it on the target machine.

I chose to partition this way and use lodlin rather than lilo because

1. The main partition (6 gig) is to large to accomodate booting with lilo alone and would have required an additional smaller partition located within the first 1024 disk addresses.
2. In the event that one drive fails catastrophically, the system must continue to run and be bootable with minimum effort and NO data loss.
   • If either hard drive fails, the boot will abort, and the rescue system will run. Examination of the screen message or /dosx/linux/raidstat.ro will tell the operator the status of the failed array.
   • If hda fails, the dos partition on hdc must be made 'active' and the bios must recognize hdc as the boot device or it must be physically be moved to the hda position by re-cableing. The raid system can then be made active again by removing the failed drive and issuing:

              "/sbin/mkraid /etc/raid1.conf -f --only-superblock"
     

     to rebuild the remaining superblock.
   • Once this is done, then

             mdadd -ar
     

   • Examine the status of the array to verify that everything is OK then replace the good array reference with the current status until the failed disk can be repaired or replaced.

             cat /proc/mdstat | grep md0 > /dosa/linux/raidgood.ref
     
             shutdown -r now
     

     to do a clean reboot, and the system is up again.