Re: [PATCH net 0/2] vxlan: Set a large MTU on ovs-created vxlan devices

[David Wragg <david-1SEAoVOfG6VEzL6FDj/jAg@public.gmane.org>]

Jesse Gross <jesse@kernel.org> writes:
> On Wed, Jan 6, 2016 at 3:25 PM, David Wragg <david@weave.works> wrote:
>> I'm certainly open to suggestions of better ways to solve the problem.
>
> One option is to simply set the MTU on the device from userspace.

If that worked I wouldn't be submitting a patch.

The MTU value of 1500 is not merely the default.  It is also the maximum
allowed for a vxlan netdev not associated with an underlying netdev.  If
you do e.g. "ip link set dev vxlan-6784 mtu 8950", where vxlan-6784
was created by an ovs vport, it fails with EINVAL.

The first patch of the two submitted removes that limit.

> The reality is that the code you're modifying is compatibility code.
> Maybe we should make this change to preserve the old behavior or old
> callers (although, again, it should not be just for VXLAN). But no new
> features or tunnel types will be supported in this manner.

That's fine.  Naturally, the ideal from our point of view is if the
compatibility code is fully compatible, so we don't have to make changes
on our side that involve different code paths for different kernel
versions.  That's what my patches are intended to achieve.

But we can live with such changes on our side, as long as there is some
reasonable way to do so.  In the case of this vxlan MTU issue, there
doesn't seem to be one.

> New or updated userspace programs should work by simply creating and
> adding tunnel devices to OVS. That won't go through this path at all
> so you're going to need to find another approach in the near future in
> any case.

Ok.  But please try to be gentle on the poor souls who have to come up
with a single codebase that works on a range of kernel versions going
back a few years.

David
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev