How to ignore OSPF MTU checking?

[Ondrej Zajicek]

On Mon, Mar 25, 2019 at 07:23:02AM +0000, Kenth Eriksson wrote:
> > On Fri, Mar 22, 2019 at 09:24:58PM +0100, Piotr Marciniak wrote:
> > > Cisco switches require this for proper OSPF operation, if MTU is
> > > set
> > > different on equipment. But afaik it does not hit Bird. We have
> > > such
> > > sessions between Cisco and Bird. And works fine without this
> > > setting on
> > > Bird. But maybe therr is scenario we have never met but Kenth
> > > did?Best
> > > wishes,Peter
> > 
> > We have a bug in older versions of BIRD that it does not verify that
> > MTU
> > of both sides is the same. We fixed that in the latest versions.
> > 
> 
> How does bird handle transmission of OSPF packets packets larger than
> the link MTU? Does it rely on IP fragmentation? Or can it split into
> multiple protocol packets?

BIRD splits OSPF traffic to packets smaller than link MTU when possible.
If it is not possible (e.g. one LSA larger that MTU), larger packet is
used and we rely on OS to do fragmentation. We also suppose that OS
assembles incoming packets transparently. AFAIK that happens in Linux,
not sure about BSD.

> What is the expected bird behaviour if the link mtu is different of the
> two sides?

Expected behavior (specified by OSPF RFC) is to forbid an association to
be established.

-- 
Elen sila lumenn' omentielvo

Ondrej 'Santiago' Zajicek (email: santiago at crfreenet.org)
OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net)
"To err is human -- to blame it on a computer is even more so."