[vox-tech] Software IDE Root RAID on Linux

[Stephen M. Helms]

msimons@moria.simons-clan.com wrote:
> 
> On Wed, Mar 13, 2002 at 05:11:31PM -0800, Jeff Newmiller wrote:
> > On Wed, 13 Mar 2002, Bill Broadley wrote:
> > > On Wed, Mar 13, 2002 at 10:23:40AM -0800, Stephen M. Helms wrote:
> > > > This makes perfect sense on a production machine, not needed in my
> > > > case.  I will remember for further reference for sure.
> > > >
> > > > Bill:  How do you handle swap partitions on this machine?
> > >
> > > Normal swap partitions, as recommended, it makes sense since you don't
> > > need the overhead of the raid layer and swap already will make use
> > > of multiple disks for increased performance.
> >
> > Not a RAID user, but this doesn't make sense to me.  If part of my swap
> > goes south, how can my applications continue to run?  I think it depends
> > on whether you require continuous reliable operation or are only willing
> > to pay for safe data and nevermind a crash related to hardware failure.
> 
>   I agree, swap on raid should be a benefit to stability in drive failure
> cases.
> 
>   Any running applications that have pages swapped out to a failed drive
> _will_ have to crash when they attempt to draw that page back in to
> memory for use.  I suspect they would probably receive a SIGSEGV,
> (this would be fun to test).

I also agree this would be a benefit, it just does not seem to be
supported.  I could be wrong though as the how-to is dated and I have
not set this up yet and would be using a later kernel.