[ovs-dev] [PATCH 1/5] ovn-nb: Consolidate documentation of distributed gateway ports.

Ben Pfaff blp at ovn.org
Wed Mar 18 18:45:08 UTC 2020


On Mon, Mar 16, 2020 at 02:57:25PM +0530, Numan Siddique wrote:
> On Sat, Mar 14, 2020 at 4:03 AM Ben Pfaff <blp at ovn.org> wrote:
> >
> > The documentation of distributed gateway ports was in two places: one
> > place for the main documentation, another for the options.  This puts
> > it all in one place to make it easier to follow.
> >
> > Signed-off-by: Ben Pfaff <blp at ovn.org>

...

> The option - "reside-on-redirect-chassis" applies to normal logical router
> ports  and not to a distributed gateway port.
> 
> This option is considered by ovn-northd only if the
>    - router port's logical switch has a localnet port
>    - The router to which this port is connected has a distributed
> gateway port in it.
> 
> (The reply here some more details  -
> https://mail.openvswitch.org/pipermail/ovs-dev/2020-March/368595.html)

OK, I have spent a lot of time reading code and documentation and commit
message nd now I think I understand, mostly.  I rewrote everything
entirely for v2, which I will send soon.  Please do review it.  I'm not
going to push it without your feedback.

The following may be more of a question for Ankur.  I don't fully
understand options:redirect-type=bridged.  As I understand it, the
problem that this (and reside-on-redirect-chassis) is intended to solve
is an MTU problem.  I see how reside-on-redirect-chassis solves the MTU
problem, by avoiding tunneling entirely.  For redirect-type, it seems
like the MTU problem would still be there for east-west communication,
since that still goes through tunnels.  Reducing the VM's MTU would
solve that problem, but it seems like it would also solve the problem
that redirect-type/reside-on-redirect-chassis solves.

Thanks,

Ben.


More information about the dev mailing list