Re: G'damn raidframe... (was: RaidFrame with unequal size replacements)

[Chris Ross <cross+netbsd%distal.com@localhost>]

On Aug 1, 2009, at 4:04 PM, Manuel Bouyer wrote:
> > > I think you forgot to
> > > raidctl -A yes raid0
> >
> >  I don't think that was the problem.  At least, the raid array did
> > configure itself, because I booted off of it.
>
> But you said you changed the name by disabling autoconf. Did you
> reenable autoconf after that ?

  I'm sure I did.  I've been doing so much "raidctl -A no raid0"  
"raidctl -A root raid0" over the last few days, I must admit I don't  
remember it all in great detail.  ;-)

> I never needed to do it, but I didn't change the raidframe's name ...

  That's true.  I did, at some point, accomplish changing the name,  
as suggested, by setting it to not autoconfigure, then manually  
configuring it when no other raidframe was present.  But, that's no  
longer the issue.  The problem I had was after the rename process, I'm  
99% sure.

> If you disable autoconf, you need to reenable it at some point.

  At the moment, I'd reenabled autoconf (root) to bring the system  
up, added the new disk as a hot spare, and done a rebuild onto it by  
failing the magic "component1" component.  I now have:

Components:
           /dev/sd0a: optimal
          component1: spared
Spares:
           /dev/sd1a: used_spare


  Which is as described in the documentation.  The statement there  
was that just rebooting at this point would cause everything to whack  
back into place, with the two actual disks in the array as compoents,  
and no spares.  But, in my first attempt, this didn't happen.  I  
suspect I did something else wrong, which I hope I didn't again, but  
just wanted a sanity check before rebooting at this point.

  I'm away from the machine this evening, so won't reboot it until  
tomorrow afternoon (GMT-4), but was just looking for a "yeah, that  
should work".  :-)

  Thanks.

                 - Chris