Alan Barrett writes:
> On Thu, 18 Jan 2001, Manuel Bouyer wrote:
> > No, if there is a failed disk raidctl -c will fail, and the boot will stop.

No.. that should *NOT* be the case...  'raidctl -c' was designed to succeed
as long as there were enough components there to recover all the data... 
if it's not doing that, there is a but somewhere that needs to be fixed...

> OK, you are right.  raidctl could probably be taught to DTRT here;
> perhaps via a new command line option to make it treat errors in the
> same way that autoconfiguration treats errors (ignore missing disks,
> mark disks as failed if their modification count is too low but
> everything else is OK).

That is what it's supposed to be doing right now... if it does 
something different, please let me know...  (I just tested this on my 
test box, and it boots just fine with 'raidctl -c' on a RAID5 set with a 
failed component...)

> BTW, autoconfiguration doesn't work with recursive RAID-on-RAID, so I
> don't think we can remove the boot-time "raidctl -c" commands.

I just need to find time to make the auto-detect stuff a little recursive  :)

Later...

Greg Oster