[rhos-list] Folsom test with Cinder and Netapp iScsi driver.

[Eric Harney]

On Wed, 2013-03-06 at 14:04 -0800, Chris Wright wrote:
> * Thomas Oulevey (thomas.oulevey at cern.ch) wrote:
> > I was wondering if someone has some experience with Cinder and the iScsi
> > Netapp driver.
> 
> I haven't tried it.

I also haven't tried the Netapp driver.

> 
> > We managed to use it through the OnCommand API and to
> > create/attach/detach/reattach volume.
> > 
> > 2 questions:
> > - The snapshots seem broken. It is not clear if it is available or not. Any
> > idea ? Success ?
> >  The API seems to handle the request but we could not make it to work.
> >  https://wiki.openstack.org/wiki/CinderSupportMatrix
> > - When you attach a volume it seems the device name choice is not respected:
> >  volume end up all the time with /dev/vdb. (If vdb exist it will go to the
> > next available vdc in this case) but
> >  the cli and horizon show the user choice (/dev/vdd for example).
> >  Even the dumpxml show the user chosen device.
> 
> Good questions above.  I don't have answers, perhaps Eric (Cc'd) does.
> 
> > I didn't find a bugzilla but there is a discussion here :
> > https://bugs.launchpad.net/nova/+bug/1004328
> > The solution seems more a workaround e.g using a SN based naming.
> > 
> > Let me know if I should open a bugzilla issue where I can attached the
> > details, logs, etc....
> 
> Yes, open a bug to help track.  We can always close it if it turns out to
> be some simple configuration type error.
> 

I went ahead and opened a doc bug for the device node issue to start,
because my impression is also that it doesn't behave as expected.  I
don't know all of the specifics about how this works since it's more in
the Nova/virt area, but I do think there may be some unexpected
limitation here that we need to look at.

https://bugzilla.redhat.com/show_bug.cgi?id=918809

Thanks for pointing this out.

> thanks,
> -chris