[ovirt-users] newbee questions

[Dan Kenigsberg]

On Mon, May 12, 2014 at 06:14:40PM +0200, John Smith wrote:
> On Mon, May 12, 2014 at 5:39 PM, Dan Kenigsberg <danken at redhat.com> wrote:
> >
> > A big limitation of the macvtap approach is that it would let you
> > connect only a single VM to your WiFi. Is that fine by you?
> >
> I wasnt aware of that limitation. No, a single VM limitation would not
> be sufficient. I was assuming you could simply do something like this
> :
> 
> ip link add link wlp2s0 name macvtap0 type macvtap
> ip link add link wlp2s0 name macvtap1 type macvtap
> ip link add link wlp2s0 name macvtap2 type macvtap
> 
> And then assign individual 'interfaces' macvtap0, macvtap1 and
> macvtap2 to different VM's. Apparently it doesnt work that way.

Maybe it does - please try or wait for someone in the know (such as mst) to
explain. The thing is that afaik
http://libvirt.org/formatnetwork.html#examplesDirect creates the macvtap
devices for you, and it creates one device per one virtual function.

> 
> 
> >
> > Yes, defining a libvirt network requires using
> >
> >   virsh net-define <xmlfile>
> >
> > (you may need to pass the not-so-secret vdsm at ovirt/shibboleth password
> > to tinker with libvirt, which is better done on non-production setup)
> >
> >   virsh net-dumpxml <net-name>
> >
> > could show you what's already defined, but I don't expect you to have
> > interesting networks as of yet.
> >
> Thanks, ill start fooling around with virsh net-define and virsh
> net-dumpxml then.
> 
> 
> >
> > It may not be easy, but I'd be grateful if you report here on what you
> > will have accomplished.
> >
> No problem. Although after having read your last message, I think I
> may be better off using (and start with looking at) your NAT-based
> network / vdsm-hook-extnet approach.

NB: Even if it is possible to define a libvirt network with several
pre-created macvtaps, you'd need something like vdsm-hook-extnet to
convince oVirt to use your network instead of a bridge.

Dan.