[dm-devel] dm-mirror: do not degrade the mirror on discard error

[Mike Snitzer]

On Wed, Feb 18 2015 at 12:54pm -0500,
James Bottomley <James.Bottomley at HansenPartnership.com> wrote:

> On Wed, 2015-02-18 at 12:10 -0500, Mike Snitzer wrote:
> > 
> > Let's please set this thread to one side.  I already offered my thoughts
> > in this thread, see: 
> > https://www.redhat.com/archives/dm-devel/2015-February/msg00076.html
> > 
> > I'm not interested in wiring up the actual error return for dm-mirror's
> > benefit.  If/when other dm-io consumers have a need to differentiate
> > between error codes we'll fix dm-io as needed.
> 
> Yep, concur ... the point I've been obviously failing to make is that
> all the world is not an SSD.

I agree, and am aware.  The layers that have needed to worry about
properly differentiating between these higher level returns from SCSI do
(e.g. dm-mpath).  But it is pretty rare for bio-based DM targets to
differentiate between error codes.  Not saying that is right, just that
is how things stand.  Could be there is room for improvement in places.