[c-nsp] MTU size, fragmentation and drops

Andrew Miehs andrew at 2sheds.de
Sat Apr 16 01:11:41 EDT 2016

The PE was sending out packets bigger than CE could accept, due to
your settings and/or the link characteristics.
Its like trying to fit a domestic hose attachment (MTU1500) to a
firehose (MTU9000) without an adapter. Its just not going to work, and
water (PACKETS) spray out the side everywhere, and nothing goes
through the domestic hose nozzle.

There is no chance of the CE sending back an ICMP fragmentation needed
message as the frame it is supposed to ingest never enters into the
router, cause the 'ingress' port is too small to allow it to fit. As
Gert said, L2 vs L3.

On Fri, Apr 15, 2016 at 8:35 PM, Mikael Abrahamsson <swmike at swm.pp.se> wrote:
> On Fri, 15 Apr 2016, CiscoNSP List wrote:
>> My question is why?  i.e. Why would reducing our PE's Ints MTU size to
>> 1500 "allow" packets above 1500bytes to pass fragmented, but at 9100, they
>> were dropped?
> There are two values here, MTU and MRU.
> Some devices configured for 1500 MTU are capable of receiving 9100
> unfragmented packet, but will not transmit more than 1500 bytes (thus
> fragmenting before sending).
> So most likely, the CE had MTU and MRU set to 1500, and since the 9100 sized
> unfragmented packet sent from the PE to the CE was higher than the MRU, it
> was dropped.
> If there were some L2 switches in between the PE and CE, then it/they might
> have dropped it.
> --
> Mikael Abrahamsson    email: swmike at swm.pp.se
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/

More information about the cisco-nsp mailing list