ata0-raid oddness.

Søren Schmidt sos at
Sat Jan 31 03:29:19 PST 2004

David O'Brien wrote:
> On Sat, Jan 24, 2004 at 12:27:07AM +0100, Dag-Erling Sm?rgrav wrote:
>>Lanny Baron <lnb at FreeBSDsystems.COM> writes:
>>>That is correct. ad0 and ad1 are subdisks of respective ar*
>>it is *not* correct - ad0 and ad1 should not be shown when they are
>>members of an active array, as any attempt to partition and label them
>>directly is likely to corrupt the array.
> older ATA didn't show the members of an active array.  ATAng started
> showing them. :-(  I asked sos about it and he said they'ed be exposed
> for a while until he finished some things he was working on.
> Maybe sos can update us on the state of things and the plans.

Real old ATA (4.x) showed them and when they left in preATAng there was 
lots of complaining. I see that now we are back to the old (initial) 
behavior complains are showing up again :/

There are problems like what to do with disks that contains a valid RAID 
config but that RAID cannot be completed and used, how are such disks 
supposed to show up ?

Anyhow the ataraid code was about to change, but then hope started to 
show again that we could loose ccd/vinum/raidframe/ataraid and get one 
proper GEOM implementation, that would make life perfect (about RAID at 
least) so I've put further work on ataraid on the backburner since I 
dont want duplicate work in there (we have PLENTY of that already)....

However since some of our worst armchair generals and whiners are 
involved in this, I dont expect an outcome soon, if any at all ;)


More information about the freebsd-current mailing list