Hi Erik,
On Apr 8, 2005, at 19:41, Erik Nordmark wrote:
Brian Haberman wrote:
o An anycast address MAY be used as the source address of an IPv6
packet.
o An anycast address MAY be assigned to an IPv6 host.
This change will allow users to operate IPv6 anycast services in the same
manner in which they do today with IPv4 anycast.
If we do this, shouldn't we also list the issues one needs to cope with?
For instance, when anycast is used as a source address, ICMP errors might no longer be delivered to the sender (they might be delivered to some other host that uses the same anycast address). This means that ICMP path mtu discovery might not work. As a result, I think it would make sense to say that when an anycast address is used as source address, the sender SHOULD limit the IPv6 packets to 1280 bytes.
It might also make sense to note that the lack of ICMP errors means that debugging tools like ping and traceroute might not work when an anycast address is used as source address.
I tend to think that it is worthwhile to list some of these issues in the spec. These
are not intuitive problems that people will recognize without some guidance.
Regards, Brian
smime.p7s
Description: S/MIME cryptographic signature
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
