[EMAIL PROTECTED] wrote:
>
> >>If there is enough of the packet included to translate the ICMP
> >>then it is better to both adjust the path MTU and do the translation.
> >>If you only adjust the path MTU it will take another packet drop
> >>(with an ICMPv6 packet too big from the tunnel entrypoint) before the
> >>IPv6 sender sees the path MTU. If you do both (when the ICMPv4 packet
> >>included sufficient headers) then a single packet drop is sufficient
> >>to notify the IPv6 sender of the path MTU.
> > I guess we are talking about the same behavior with different wording.
>
> oops, you right.
> "translation" - send ICMPv6 as soon as B gets ICMPv4
> normal behavior - learn PMTU, and then send ICMPv6 on next bigger packet
>
> if you don't turn on DF bit on outer IPv4 header, you can avoid this
> whole translation thing for too big messages, btw.
Yes. If you remember, RFC 2529 (6over4) says
However, the IPv4 "do not
fragment" bit MUST NOT be set in the encapsulating IPv4 header.
but we agreed in a subsequent discussion that this should be downgraded
to SHOULD NOT, and RFC 3056 (6to4) has SHOULD NOT.
RFC 2893 doesn't make a general recommendation - maybe it should?
Brian
--
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Brian E Carpenter
Distinguished Engineer, Internet Standards & Technology, IBM
On assignment for IBM at http://www.iCAIR.org
Board Chairman, Internet Society http://www.isoc.org
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
