[Discuss] file system checksums

[blu at nedharvey.com (Edward Ned Harvey)]

> From: discuss-bounces+blu=nedharvey.com at blu.org [mailto:discuss-
> bounces+blu=nedharvey.com at blu.org] On Behalf Of Tom Metro
> 
> > Hint: RAID does nothing for data integrity or consistency.
> 
> Huh? The checksums permit detecting that the two devices are
> inconsistent, and the duplication lets you repair the copy with the
> failing checksum.

Does linux raid (mdadm) perform checksumming?  I've never heard of that
before...  Most raids don't...  (ZFS raid and btrfs raid are the first time
I'm aware of a raid system providing checksumming...)  Most raids (let's
take for example, an LSI hardware raid card) just write redundant copies of
the data...  And when it comes time to read it back...  It just reads one
side of the data (doesn't read the redundant copy) and assumes the data is
good.  After all, the data passed the on-device FEC, didn't it?  Of course
it did...  Or at least the on-device FEC failed to indicate a failure...
It's that whole "trust your hardware not to make an error" thing...
Calculators never make mistakes.

I think that's what RP is suggesting.  He's never heard of it before either,
apparently.  (Linux raid, particularly md, performing checksumming)