Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 22 Jul 2007 20:09:43 +0200
From:      Sten Daniel Soersdal <netslists@gmail.com>
To:        karels@karels.net
Cc:        freebsd-net@freebsd.org, Artyom Viklenko <artem@aws-net.org.ua>, Artem Belevich <fbsdlist@src.cx>, Julian Elischer <julian@elischer.org>
Subject:   Re: 6.2 mtu now limits size of incomming packet
Message-ID:  <46A39D67.5030900@gmail.com>
In-Reply-To: <200707211702.l6LH2ukt039317@redrock.karels.net>
References:  <200707211702.l6LH2ukt039317@redrock.karels.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Mike Karels wrote:
>> Any two hosts, connected to single Layer2 network MUST use
>> same MTU. Any other cases lead to hard-to-solve problems.
> 
> I'd have to disagree.  In fact, I'd say that any two hosts on the
> same L2 network must use the same MRU.  In particular, if a host
> choses to use a lower MTU, if that also lowers the MRU, *that* is
> the cause of interoperability problems.
> 
> David DeSimone <fox@verio.net> wrote:
> } You are correct about misconfigured networks.  In my experience,
> } the only reason to ever reduce the MTU is to work around a problem
> } discovered in someone else's network (not my local segment).  Fixing
> } the problem by getting someone else to fix their network is generally
> } too hard.  If MTU == MRU was forced behavior, the viability of this
> } workaround would be removed, one less tool in the toolbag, so to speak.
> 
> Exactly.  In our local labs, we also reduce the MTU to test PMTU discovery.
> Requiring MRU == MTU makes this more difficult.  True, it's a contrived
> situation, but as you say, one less tool in the toolbag.
> 
> 		Mike

To leave this tool in the toolbag one could set interface mtu to the 
upperbounds of what one needs (e.g. 1500) and use pr route --mtu setting 
to work around those broken networks. That shouldn't break the inbound 
oversized frames (frames smaller than 1500) but lower the outbound mtu.

-- 
Sten Daniel Soersdal



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?46A39D67.5030900>