From: Olafur Gudmundsson <[EMAIL PROTECTED]> Subject: Re: ENDS0 > I discussed this with the IPng WG chairs yesterday, and > one of them suggested that the message-size draft actually mandate > EDNS0 for hosts with IPv6 installed. This is a stronger requirement > than I was willing to make, but the other chair suggested to > ask the working group if they thing this is a good idea. Just for clarification. Does this mean dropping my proposal? Or pick up both and put them into a draft of IPv6 host requirement? > Discussions on if EDNS0 should be mandated for IPv6 capable hosts > should take place here on ipng mailing list. OK. I would like to propose as follows: (1) DNS resolvers MUST implement EDNS0 if they support IPv6 transport or AAAA/A6. Such resolvers SHOULD specify 2048 (*1) bytes as buffer size to servers. (This requires larger IPv6 reassemble buffer than the minimum, ie 1500.) # (*1) need to discuss an appropriate size (2) DNS servers MUST implement EDNS if they support IPv6 transport or AAAA/A6. They MUST assume client's buffer size is 1024 (*2) if clients doesn't specify their buffer size by ENDS0. # (*2) 1024 comes from the IPv6 minimum MTU, 1280 = 256 + 1024. --Kazu -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------