[dm-devel] Status of resizing/growing dm-multipath devices on the fly

[Andrew Patterson]

On Wed, 2008-08-06 at 12:36 -0700, Mike Anderson wrote:
> Pasi K?rkk?inen <pasik at iki.fi> wrote:
> > On Wed, Aug 06, 2008 at 03:48:12PM +0200, Domenico Viggiani wrote:
> > > * Pasi Kärkkäinen wrote:
> > > > 
> > > > Is it possible to resize (grow) dm-multipath devices on the 
> > > > fly nowadays?
> > > > 
> > > > I googled and found some discussions about the subject, but 
> > > > the conclusion seemed to be it's not possible.. that was a 
> > > > while ago, so I was wondering if this has been fixed/implemented..
> > > > 
> > > > Using LVM and adding another new LUN/PV is not an option 
> > > > always.. it's a lot easier to manage the whole thing if it's 
> > > > possible to resize/grow existing LUNs on the fly.
> > > 
> > > Nowadays all disk-arrays make online LUN extension a breeze but
> > > unfortunately it seems that dm-multipath is still not able to see new size
> > > without downtime:
> > >  http://www.redhat.com/archives/dm-devel/2007-August/msg00205.html
> > > Look also at this recent thread on RHEL5 mailing-list where you can find
> > > also a post from a Red Hat representative:
> > >  http://www.redhat.com/archives/rhelv5-list/2008-July/msg00267.html
> > > 
> > > Sure, you can just add a new LUN, pvcreate and vgextend but it is not a
> > > viable solution because it causes LUN proliferation.
> > > 
> > > Sincerely, I'm still subscribed to this mailing-list only to see if someone
> > > solve this SHAME!
> > > It's a feature that Linux really needs.
> > > 
> > > Hope someone pick this cry of pain!
> > > 
> > 
> > Thanks for the reply.
> > 
> > Does someone know what's the actual problem in this? meaning why it hasn't
> > been done yet.. 
> 
> It has to do with the block layer not using the new size value while there
> are openers.
> 
> Andrew submitted a patch series (url below) a while ago to try and address
> this issue, but I do not know the status. I added Andrew to the cc.
> 
> http://thread.gmane.org/gmane.linux.scsi/41623
> 

The above patchset works with SCSI devices. I have tested it with single
SCSI devices under dm/LVM, i.e., a PV is on a single SCSI LUN with only
one path to it. It has not yet been accepted upstream due to not getting
sign-off from Al Viro. I plan on resubmitting soon to add cciss support.
Hopefully Al will getting around to reviewing it sometime this century.
I have heard rumors that Al has some all-encompassing solution for this
problem but hasn't got around to implementing it.  I'll test my new
patches with multi-path while I am at it.

Andrew