> Tunnel ingress devices already are allowed to fragment > IPv6 packets by various IETF approved tunnelling > specifications, as you have agreed.
Fragmentation of the inner packet before encapsulation is different from fragmentation of the outer packet after encapsulation. Ole was referring to the latter case, where the tunnel ingress is acting as a host when it fragments the encapsulated packet and the tunnel egress gets to reassemble. The tunnel MUST uphold the IPv6 minMTU of 1280; otherwise, it would be a degenerate link. The tunnel SHOULD allow larger packets through when it can do so without resulting in an MTU-related black hole; otherwise, it would fail to take advantage of larger path MTUs when they are available. And, the whole thing needs to work when ICMP PTBs are unavailable. This discussion is coming full circle to what we discussed earlier when we already established that tunnels MUST support the IPv6 minMTU even if some fragmentation and reassembly is necessary. Thanks - Fred [email protected] -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
