On 5/3/19 5:51 PM, John R Levine wrote:
>>> server.  I suppose you could find your DoH server by name, but if you
>>> can do that, you could equally well find your DoT or .well-known
>>> server by name and define the problem out of existence.
>>
>> I think it's best to verify by name, even if the DNS server is reached
>> through a hard-configured IP.  That's what we implemented for Knot
>> Resolver, at least.  On a related note, I'd also expect to send the name
>> as SNI by default; 8.8.8.8 was not even sending me a certificate unless
>> I sent SNI (only when using TLS 1.3 though)
>
> When I said verify by name I meant by DNS name, so the certs can be
> signed by the existing ACME protocol or whatever.

I also meant verification by DNS (host)name in certificate's CN, signed
by some commonly accepted authority.  I don't know where the
misunderstanding is.


_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to