You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Can we conclude on this discussion, if/how we could include it, or not, in the Advertising Proxy's DNS authoritative server which is used to serve query results, based on SRP-registered services, to clients. I.e. could a client use QDCOUNT > 1 (if we put in some specification effort) or not (we abandon the idea)?
The text was updated successfully, but these errors were encountered:
It doesn’t seem like we came to a conclusion. I think if we want to allow
this, the right approach is to only allow it for queries on the same name,
and to make it optional.
The problem is that in order to not cause harm, we need resolvers that
don’t support it to respond with just one answer, rather than an error. But
we can’t be sure that all resolvers that might exist in this context would
do that. So there is the risk that a resolver might respond with some
error, forcing the client to send two more messages, or not respond at all,
forcing the client to retry multiple times and then probe.
We could argue that using a resolver that doesn’t support whatever behavior
we specify is an operational error. If we don’t require that, I think we
have to forbid QDCOUNT>1.
Op zo 5 nov 2023 om 18:47 schreef Esko Dijk ***@***.***>
There was a proposal and long discussion on QDCOUNT > 1 in the thread started here: https://mailarchive.ietf.org/arch/msg/dnssd/OpEsujhoSFgC_dXBU6kTd3gtqR0/
Can we conclude on this discussion, if/how we could include it, or not, in the Advertising Proxy's DNS authoritative server which is used to serve query results, based on SRP-registered services, to clients. I.e. could a client use QDCOUNT > 1 (if we put in some specification effort) or not (we abandon the idea)?
The text was updated successfully, but these errors were encountered: