Re: [Add] TTL of resolver.arpa

I think there are some valid concerns here around clients mistakenly
applying resolver.arpa info when changing to different networks/resolvers.

But I think mandating some never-cache policy is way too strong and not the
right solution.  Many clients already do stuff like clear the cache on
network change or DNS config change, so such a policy is unnecessary for
them.  And when the network and resolver haven't changed, it is still
undesirable for the client to poll that resolver more frequently than the
TTL, so using their cache implementations is the obvious way for clients to
implement this desired behavior.

On Wed, Dec 22, 2021 at 9:45 AM Daniel Migault <mglt.ietf@gmail.com> wrote:

> Hi,
>
> I am wondering if some additional text is not needed regarding the TTL of
> the _dns.resolver.arpa RRset. As resolver.arpa is not owned by anyone, this
> information should not be cached. If one device is changing network for
> example, we should make sure the mobile will not consider the resolver.arpa
> response performed on a previous network. Similarly, when a dns client
> performs simultaneous discovery on different resolvers. Should we
> recommend/mandate the DNS client to set this TTL to 0 and not cache the
> response ?
>
> _dns.resolver.arpa  7200  IN SVCB 1 doh.example.net (
>         alpn=h2 dohpath=/dns-query{?dns} )
>
> Yours,
> Daniel
> --
> Daniel Migault
> Ericsson
> --
> Add mailing list
> Add@ietf.org
> https://www.ietf.org/mailman/listinfo/add
>