[CentOS-docs] Updated How to Setup a Software RAID on CentOS 5

Thu Apr 30 15:34:38 UTC 2009
R P Herrold <herrold at centos.org>

On Thu, 30 Apr 2009, Marcus Moeller wrote:

> I personally cannot agree on that. Syncing /boot 'manually' adds much
> more complexibility, unless you add scripts that automate the process.


/etc/yum.conf contains a:
/etc/yum-kernel.conf  does not

 	trivially wrappable in a script

# mount -n -w /boot
# yum -c /etc/yum-kernel.conf update kernel\*
#  mount -n -r /boot

# [ ! -d /mnt/boot ] && mkdir /mnt/boot
# mount -w /mnt/boot /dev/sdbX   (/boot1)
# rsync -a /boot/. /mnt/boot/.
# umount /mnt/boot

A copy and paste of the first stanza (zero index is our convention 
here) with one edit for /boot's
 	root (hd0,0)
to make it:
 	root (hd1,0)
finishes the job

> I also wonder what's wrong in having /boot on a RAID1 partition,
> because rescue mode should offer all necessary tools to mount raid
> partitions. Instead I would suggest to add a separate page with common
> recovery scenarios.

No rescue media may be available, or the host may lack a drive 
to boot it from; my method, and an alternate 'fallback' boot 
stanza in grub.conf

      # Fallback to the second entry.
      fallback 1

and one can recover with nothing more than a grub system 

-- Russ herrold