Re: [Doh] meta qtypes

[Patrick McManus <[e-mail]>]

Section 4 has spme rules about generating http freshness lifetimes from dns
responses. Is there a text change to that you are proposing?

On Mar 18, 2018 7:08 PM, "Miek Gieben" <[e-mail]> wrote:

> [ Quoting <[e-mail]> in "Re: [Doh] meta qtypes..." ]
>
>> Hi - DoH itself doesn't require axfr/etc anywhere that DNS would not (I'm
>> not aware of whether it does or not) - the text is just meant to give the
>> HTTP considerations when it does carry larger DNS responses such as that.
>>
>> WRT HTTP caching - section 4 of DoH talks about how HTTP freshness
>> lifetimes should be set for DoH responses. Is there something different
>> here?
>>
>
> This HTTP caching opens up a pretty big can of worms on what to set the
> cache
> TTL to.
>
> If we consider HTTP/2 *just* the transport than it should not be concerned
> with
> caching (i.e. TCP doesn't cache).
>
> If we do want to cache on this layer a server implementing DoH needs to
> inspect the DNS packet, it should (among other things), look for a SOA
> record to get a TTL or MINTTL, or get the minimum TTL of all the records in
> the message, or look for RRSIG expiration times and distill a minimum TTL
> from those.
>
> I think this draft should say "MUST not cache these responses".
>
> /Miek
>
> --
> Miek Gieben
>