[ovs-dev] [ovn-controller-vtep V2 6/6] ovn-controller-vtep: Add pipeline module.

[Alex Wang]

On Mon, Jul 13, 2015 at 9:24 AM, Russell Bryant <rbryant at redhat.com> wrote:

> On 07/05/2015 01:38 AM, Alex Wang wrote:
> > This commit adds the pipeline module to ovn-controller-vtep.  The
> > module will scan through the Binding table, and create the
> > Ucast_Macs_Remote in vtep database for each logical port's MACs in
> > the same logical datapath (same as logical switch for vtep).  Also
> > the module will create the physical locators to each HV chassis in
> > the same logical datapath.
> >
> > TODO:
> >
> > - my understanding is that the ovn-controller-vtep should collect
> >   all 'Ucast_Macs_Local's from vtep database and update them to
> >   the ovnsb.  However, currently, it is not clearly defined in
> >   ovnsb schema about how to do this.  Particularly, there could be
> >   logical ports from different chassis gateway ports attached to
> >   the same logical datapath.  However, we cannot tell which logical
> >   port the MAC of a 'Ucast_Macs_Local' entry belongs to.  So I think
> >   this definitely deserves more discussion.
>
> Interesting.  It seems like the set of MACs mapped to a gateway should
> be reflected in the Pipeline table, but ovn-northd currently owns that
> table.  Maybe this info should be in the Gateway table, and ovn-northd
> should reflect it in Pipeline?  I see that ovn-sb already looks at
> gateways as a logical port, so I think it would just look like other
> logical port rules.
>
>
Hey Russell,

Thx for the comments, actually, I recently realized that we can simply the
gateway part of ovn-sb schema based on the nature of vtep schema.  Also,
there is apparently a lot of complications in updating the vtep schema info
(like the learned macs) to ovn-sb or even ovn-nb.  I have lot of thoughts
will
meet Ben, maybe Justin in person this afternoon and discuss.  Let me
update more after that~.

Thanks,
Alex Wang,



> --
> Russell Bryant
>