[Discuss] On Btrfs raid and odd-count disks

[Richard Pieri]

On 4/12/2013 10:58 AM, Dan Ritter wrote:
> On Fri, Apr 12, 2013 at 10:51:21AM -0400, Derek Atkins wrote:
> mdadm has a sort of scrub facility available, in which it reads
> all the bits -- see /sys/block/$array/md/sync_action 
> 
> Most HW RAID controllers have something similar available.

RAID systems have rebuild or resynchronization facilities. This is quite
different from scrubbing a checksum file system. Take a simple mirrored
pair of devices. Say that there are bit errors in different places on
both devices. The RAID system can't know which blocks have good data and
which have bad data, so a rebuild will copy one device to the other, bad
data along with the good.

Checksum file systems protect against this. The checksums give the
storage system the capability of detecting bad data, regardless of
cause, and attempt repair if sufficient good replicas exist.

-- 
Rich P.