On Thu, 2004-09-02 at 06:39, Christoph Kaegi wrote:
> Quoting Brian A. Seklecki (lavalamp@spiritual-machines.org):
> > 

> > Right now I'm looking for people to help test and proofread the new
> > material.
> 
> I'll review the document and send my feedback in your direction.

Thank you.

> 
> I'll also test it by setting up RAID-1 on a 2.0Beta machine

There were two conflicting chains of thought with regard to allocating
RAID volumes.

The 1st is to partition several smaller slices of FS_RAID on your
component disks, then establish several RAIDFrames, each composed of two
slices, each with *one* FFS/4.2BSD file system occupying the entire a:,
c: or d: slice.

The 2nd is to dedicate the entire component volume disk to FS_RAID,
create one large RAIDFrame volume, then partition several 4.2BSD/FFS
slices within that pseudo-volume.

Honestly, I couldn't see an advantage to either so I went with the
more-common #2.  The only advantage that I could imagine would be a
small component sector size, but since RAIDFrame seems to use a
512byte/sector block size regardless....

Perhaps from a redundancy stand point, if a component volume develops
errors on low-level sectors, it would affect one RAID volume and not
another, but with modern disks, those read/write re-trys are still going
to take all the other volumes residing on that physical component
offline temporarily.

Thoughts anyone?

~lava

> for the first time.
> 
> Until Later
> 
> Chris