On Fri, Jul 3, 2020 at 10:59 AM John Levine <jo...@taugh.com> wrote: > In article < > cah1icirmlslmohchqcvqis6ra0myk40ejsdm_b5pmxagxrn...@mail.gmail.com> you > write: > >There are a whole bunch of unused bits in the core element of the OPT RR > >(the place where the DO bit exists). That would be an excellent (IMHO) > >place to signal the situation here (partial glue truncation). > > > >Such a bit would hopefully disambiguate the cases where TC=1 is set, > >allowing for graceful handling (try to use the available glue, prepare for > >the failure case and retry over TCP if it does fail). > > > >Thoughts? > > I think it would be easier to clarify the existing rules and encourage > people to follow them rather than inventing new flags that are > unlikely ever to be implemented. > > For the particular case of partial glue on referrals, so long as the > packet has all of the answer records, it's never wrong to go ahead and > use the glue it's got. What would a flag do? >
It makes clear the difference between implied and inferred. The flag clearly indicates that some glue which would otherwise be considered necessary has not been sent. It simplifies the implementation on the receiver's side: no flag, all glue was sent; flag, some glue is missing. The recipient never has to determine whether there should be glue or not, including whether the sender is a busted implementation or zone. The recipient has a clear path of action, if it is unable to contact any of the name servers from the referral due to missing glue: ask again (over TCP). Additionally, the flag would signal compliance with updated 1035, for starters. The uptake on the clarified 1035 won't be instantaneous or universal, so it makes it possible to correctly infer whether a server is doing things the new way or not. Also, it would facilitate lower effort in figuring out if a TC is referral related or not, which could be important for implementers/operators doing large scale stuff. And lastly, it would strongly encourage anyone not currently doing the partial glue usage thing, to do so. Most importantly, this is all about interoperability, including not just the wire format but the operational signaling. Not everyone will implement things the same way, but at least they will have a signal about how the other party is doing what they are doing. Brian
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop