Re: RaidFrame with unequal size replacements

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Thu, Jul 30, 2009 at 08:54:42AM +1000, George Michaelson wrote:
> A related question Manuel (or others)
>
> if 'repurposing' dk wedges on disks with existing GPT partition  
> structure how deeply do you have to 'clean' the disk to prevent it being 
> (re)detected as part of an old RAIDframe structure?
>
> the raidframe manual page for raidctl talks about failing components,  
> and adding components but not about REMOVING components as I can see.
>
> And, since a GPT label is at the back of the disk, as well as the front 
> of the disk, I'm not sure a dd if=/dev/zero of=/dev/rawdisk bs=8k count=1 
> is adequate to wipe state.

The raidframe label is at the start of the partition so I'd use the
/dev/dk* instead of the raw disk. 

>
> (or, that you even have to, given dkctl will remove wedge labels)

If dkctl really remove wedge labels then no further work should be
needed: on next boot the wedge should not be detected, so raidframe
can't find its component either. But I'm not sure dkctl will remove
the GPT label, so the partition may well be automatically detected
again (because of DKWEDGE_METHOD_GPT). 'gpt remove' would more probably
do what you want.

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--