Greg, thanks for your help.

Greg Oster wrote:-

> > 1) Remove the old raid info from the disk, or
> 
> You can nuke the component label with:
> 
>  dd if=/dev/zero of=/dev/rwd0e bs=1k skip=16 count=1
> 
> where "rwd0e" is the offending partition.

Sadly that didn't work.  I get device is read-only for
/dev/rwd0a, and if I use /dev/wd0a I get device is busy.

> > 2) Stop the kernel trying to autoconfigure a non-existent
> >    raid?
> 
> If you boot with the "askme" flag (specified with "-a" as a bootflag 
> on a number of archs) then the kernel will ask you where you'd like / 
> to be, and RAIDframe will not hijack it.

This helps a lot, thanks.  I got to a shell and from there did a
raidctl -A no and -u on each raid device.  Now I can boot cleanly!

> Alternately, you could just change the partition type in the disklabel
> from "RAID" to "unused", and it should be happy too.

That was my first thought, but none of the types are RAID.  They are
what you'd expect after a fresh install.

This should probably be fixed in the installer - I'd help if I knew
what to do!

Thanks again,

Neil.