Hi Joe, > -----Original Message----- > From: Joe Touch [mailto:[email protected]] > Sent: Monday, March 02, 2015 12:58 PM > To: Templin, Fred L; Ronald Bonica; Carlos Pignataro (cpignata) > Cc: [email protected] > Subject: Re: [Int-area] draft-ietf-intarea-gre-ipv6 > > > > On 3/2/2015 12:44 PM, Templin, Fred L wrote: > > Hi Joe, > > > >> -----Original Message----- > >> From: Joe Touch [mailto:[email protected]] > >> Sent: Monday, March 02, 2015 11:25 AM > >> To: Templin, Fred L; Ronald Bonica; Carlos Pignataro (cpignata) > >> Cc: [email protected] > >> Subject: Re: [Int-area] draft-ietf-intarea-gre-ipv6 > >> > >> Fred, > >> > >> We've had this entire discussion before. > >> > >> We should label it with an name and in the future cite by reference. > >> > >> It is out of scope until the WG decides it's important to deal with this > >> issue, IMO - and that hasn't happened yet. > > > > The GRE folks are saying that it is OK for the tunnel ingress to send PTB > > with size less than 1280. > > I'm not sure what you're getting at.
I'm getting at what they wrote. Nowhere else (except for IPv6/IPv4 translators) does it say that it is OK to send a PTB with size smaller than 1280. > Either they're saying that this is already what's deployed or they're > saying it's something new. If it is already deployed, then what happens if it is ever invoked? The document does not say. Once a source receives a PTB with a size smaller than 1280, it is supposed to start sending future packets as large as 1280 and with a fragment header included. What is the GRE ingress supposed to do then? Fragment the inner? Fragment the outer? Drop and return yet another PTB with size smaller than 1280? That's the point; it doesn't say. If the intent is to document what's already deployed, then it is only documenting one half of the behavior. The other half is what happens *after* a PTB with size smaller than 1280 is sent. On this point, the document is silent. Thanks -Fred [email protected] > The former is the purpose of the current doc. The latter is new work. > > > That forces us to consider what happens when > > a source node actually receives and acts on that message. > > It forces them to document existing behavior. Not for us to extrapolate. > > Joe _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
