Thanks for forwarding a pointer to that document. You might also see RFC5508, "NAT Behavioral Requirements for ICMP". RFC5508 includes a reference to ICMP Extensions (RFC4884) which, while rare in the wild, are probably worth describing so they are handled properly. Hairpinning of ICMP messages may also be worth pointing out to implementers (so that pinging the 'external' address works, among other things). It might be worth also pointing out that ICMP echo request/reply doesn't require special handling.
-d > -----Original Message----- > From: [email protected] [mailto:[email protected]] On Behalf > Of Steven Blake > Sent: Saturday, January 28, 2012 7:49 PM > To: nat66 > Subject: [nat66] [Fwd: I-D Action: draft-blake-nptv6-icmp-01.txt] > > Terry Moes and I have submitted this draft (actually version -01) which > describes how NPTv6 translators need to process ICMPv6 error messages. > We believe that this issue was first discovered when Terry built an > NPTv6 implementation for Linux. > > We intend to request publication as an individual submission, but first > we would really appreciate your review comments. > > > Regards, > > // Steve _______________________________________________ nat66 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nat66
