On Thu, Apr 14, 2016 at 1:05 AM, Ben Pfaff <[email protected]> wrote:
> On Mon, Apr 04, 2016 at 05:58:28AM -0700, Russell Bryant wrote:
> > This patch implements one approach to using ovn-controller to implement
> > a software l2 gateway between logical and physical networks.
> >
> > A new logical port type called "gateway" is introduced here. It is very
> > close to how localnet ports work, with the following exception:
> >
> > - A localnet port makes OVN use the physical network as the
> > transport between hypervisors instead of tunnels. A gateway port still
> > uses tunnels between all hypervisors, and packets only go to/from the
> > specified physical network as needed via the chassis the gateway port
> > is bound to.
> >
> > - A gateway port also gets bound to a chassis while a localnet port does
> > not. This binding is not done by ovn-controller. It is left as an
> > administrative function. In the case of OpenStack, the Neutron plugin
> > will do this.
> >
> > Signed-off-by: Russell Bryant <[email protected]>
>
> This seems reasonable to me. Thanks for working on it.
>
> There's a stray " * " in the middle of the second line here:
>
Oops ... I must have joined 2 lines and forgot to clean it up.
> > + * ovn-controller does not bind gateway ports itself.
> > + * Choosing a chassis * for a gateway port is left
> > + * up to an entity external to OVN. */
> > + sset_add(&all_lports, binding_rec->logical_port);
> > + add_local_datapath(local_datapaths, binding_rec);
> > } else if (binding_rec->chassis == chassis_rec) {
> > if (ctx->ovnsb_idl_txn) {
> > VLOG_INFO("Releasing lport %s from this chassis.",
>
> The treatment of the "chassis" column in the Port_Binding database is
> quite different based on the type of the port:
>
> * For localnet ports, it's always empty.
>
> * For gateway ports, ovn-northd (or whoever adds the port, I
> guess) must set the chassis correctly.
>
My thinking for this was "something external to OVN", at least for now.
For OpenStack, we could have OpenStack choose the chassis.
Letting ovn-northd do it is an interesting idea, though. It sounds worth
considering in a future iteration.
> * For other ports, ovn-controller sets (and un-sets) the
> chassis.
>
> I don't think these differences can be gleaned from the documentation.
> I think that it should be made clear in ovn-sb.xml.
>
Thanks, yes that difference definitely could be more clear.
> I guess that you may have comments from others that you should address,
> but for me:
> Acked-by: Ben Pfaff <[email protected]>
>
Thanks for the review! I didn't have any outstanding changes. Justin
asked that I wait for his review, though.
--
Russell Bryant
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
