[DNSOP] Re: draft-ietf-dnsop-3901bis-10 telechat Dnsdir review

[Tobias Fiebig <[e-mail]>]

Hello Joe,
> This is not true ("safely assume"). There is a complex graph of
> actors between stub resolvers and authority servers in the real
> world, many of which originate queries with RD=1. 
> 
> For example, ISP resolvers which forward queries to public resolvers
> with RD=1 are commonplace.  Home gateways that receive queries from
> devices within the home, and forward to other upstream resolvers with
> RD=1 following a cache miss are commonplace. These are not niche
> configurations.
> 
> I have not read your proposed changes to the text to address the
> comment from Geoff that prompted your response above, but if it is
> based on the "safe assumption" above you may want to revisit it.


The text now reads:

"Finally, when responding to recursive queries, i.e., a query with the
RD bit set <xref target="RFC1035"/>, a DNS resolver SHOULD follow..."

It is not relevant what kind of configuration sent the query. If this
was libc, a local dnsmasq, a dnsmasq on a CPE, an ISP's recursive
collecting all the queries from all CPEs and forwarding them to quad
1/8/9 etc.

It is about handling recursive queries. And if the RD bit is set, the
query is recursive, and is not being recursed by the client who sent
the query, but the recursive receiving it is supposed to recurse it.

With best regards,
Tobias

-- 
Dr.-Ing. Tobias Fiebig
T [phone]
M [e-mail]
Pronouns: he/him/his