Hi Joe, > -----Original Message----- > From: Joe Touch [mailto:[email protected]] > Sent: Friday, February 27, 2015 1:58 PM > To: Templin, Fred L; Ronald Bonica; Carlos Pignataro (cpignata) > Cc: [email protected] > Subject: Re: [Int-area] draft-ietf-intarea-gre-ipv6 > > > > On 2/27/2015 10:45 AM, Templin, Fred L wrote: > > Hi Joe, > ... > >>> Generation of atomic fragments for any reason is sufficient information > >>> for the tunnel ingress to make a determination to fragment or drop > >>> based on the size of the atomic fragment. > >> > >> The former is a direct violation of RFC2460. The latter is what > >> could/should have been done in the first place when the PTB was sent. > > > > Network fragmentation of IPv6 packets that do not include a fragment > > header is the behavior that is prohibited by the specs. Fragmentation > > of IPv6 packets that include a fragment header is already mandated > > for IPv6/IPv4 translators. So, given there is already one example of > > network fragmentation we are simply introducing another example. > > If you use the atomic fragment ID to generate an ID for the tunnel > encapsulation and fragment THAT, yes.
No, that is not correct. The ID supplied by the original source applies to the payload packet; not the delivery packet. The tunnel ingress must supply its own ID for the delivery packet, and it must not be based on the ID supplied by the original source. > If you intend to fragment the atomic arriving datagram, no. That's not > what the atomic ID is for - it is intended to provide an identifier for > the tunnel, not to allow on-path fragmentation of the original datagram. No, the ID supplied by the original source applies to the payload packet; not the delivery packet. > ... > >> It's a huge stretch to change IPv6 from "never fragment on-path" to > >> "fragment on-path". > > > > "Fragment on-path" is already mandated for IPv6/IPv4 translators. > > No; that's fragmenting at IPv4, i.e., the outer header. That's always > been allowed. Ah, here is the source of the confusion. IPv6/IPv4 translators do not use encapsulation so there is no such thing as an outer header. The translator "translates" the IPv6 header into an IPv4 header and so the packet remains as a payload packet and does not include any "outer header". So, the ID supplied by the original source applies to the payload packet even after the translator has performed v6/v4 translation. Thanks - Fred [email protected] > Joe _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
