[CLUE-Tech] RAID 1 on Linux

[Keith Hellman]

On Wed, Oct 20, 2004 at 09:28:51AM -0600, Greg Knaddison wrote:
> On Wed, 20 Oct 2004 09:19:00 -0600, Keith Hellman
> <khellman at mcprogramming.com> wrote:
> > On Tue, Oct 19, 2004 at 09:46:34PM -0700, Carl Schelin wrote:
> > 
> > > The new installation required that /boot be a regular
> > > slice rather than a raid partition. I believe that's
> > > because it's not compiled in to the kernel but a
> > > module. A regular partition will let it boot and load
> > > the module which then lets it read and mount the
> > > /dev/md partitions.
> > 
> > Could it be that your bootloader (grub, lilo, ...) can't read /boot when
> > it is a raid block device?
> > 
> > I'm not sure how much the kernel actually reads from /boot.  I'm pretty
> > sure I've set up systems that don't even *mount* /boot, but the kernel
> > (and the system) come up just fine.
> > 
> > Come to think of it, the same works AOK when you boot a box off of a
> > tftp'd kernel.  The more I think about it the more I think just the
> > bootloader uses /boot.
> 
> Depends on your bootloader, IIRC.  GRUB uses the notation (hd0,0) to
> indicate where your /boot partition is, and I believe LILO uses
> something else.  I believe this is why some people are fervently in
> favor of one or the other of hte boot loaders.
 
I think my point has been missed.  You are referring to how grub v. lilo
*refers* to a particular persistent storage area.  What I was saying is
that both grub & lilo have to be able to *read* the storage area; they
have to know how to 'parse' the filesystem in order to find the file
that should be placed into memory and jumped to (aka, the kernel).

Not being able to boot linux off of a partition which is part of a RAID
array is probably more of an issue with the bootloader's inability to
read a RAID stripe (or whatever they're called) and then be able to read
the filesystem *on* the stripe.

I would imagine that, in the past, *some* bootloader for *some*
operating system (Solaris? NT? IRIX? AIX? ...) could have probably done
this.  In truth, I dunno.

-- 
Keith Hellman                             #include <disclaimer.h>
khellman at mcprogramming.com                from disclaimer import standard
public key @ www.mcprogramming.com

"But VCs are mistaken to look for the next Microsoft, because no startup
can be the next Microsoft unless some other company is prepared to bend
over at just the right moment and be the next IBM."

-- http://www.paulgraham.com/gh.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://cluedenver.org/pipermail/clue-tech/attachments/20041020/fd9e6e2d/attachment.bin