Unfortunately -11 introduced two issues Which I've already raised on the dpriv mailing list. This is the text I sent to dpriv yesterday: ------------------------------------------------------------------------ For example, consider an authoritative server named ns0.example.com that is served by two installations (with two A records), one at 192.0.2.7 that follows this guidance, and one at 2001:db8::8 that is a legacy (cleartext port 53-only) deployment. It doesn't have two A records. It has an A and AAAA record. I know that Éric asked for a non-legacy IP example, but I don't think this makes things better. I find it very confusing, usually the server would be dual stacked so why would it do different things depending on the address family? Maybe just go v6 only, thusly? For example, consider an authoritative server named ns0.example.com that is served by two installations (with two AAAA records), one at 2001:db8::7 that follows this guidance, and one at 2001:db8::8 that is a legacy (cleartext port 53-only) deployment. A recursive client who associates state with the NS name and reaches 2001:db8::7 first will Same in 4.5: For example, if a recursive resolver can send a packet to authoritative servers from IP addresses 192.0.2.100 and 2001:db8::200, it could keep two distinct sets of per-authoritative- IP state, one for each source address it uses, if the recursive resolver knows the addresses in use. Keeping these state tables distinct for each source address makes it possible for a pooled authoritative server behind a load balancer to do a partial rollout while minimizing accidental timeouts (see Section 3.1). It seems unlikely that the load balancer would do a address family translation. Maybe: For example, if a recursive resolver can send a packet to authoritative servers from IP addresses 2001:db8::100 and 2001:db8::200, it could keep two distinct sets of per-authoritative- ------------------------------------------------------------------------