* Karl Auer: > What, exactly, is a node supposed to do when it receives a PTB < 1280 > after it has sent an ordinary packet? > > a) fragment at 1280, regardless of the returned PMTU value > > b) fragment so as to produce fragments of sizes equal to or smaller than > the returned PMTU value, but don't "remember" the returned PMTU value > > c) create a new packet that is identical to the original packet but has > a fragment header inserted (i.e., forms a fragment sequence one fragment > long, with More-Fragments=0 in the fragment header) > > d) fail to send the packet > > e) something else...? > > Only b) seems remotely useful, but it makes a mockery of the idea that a > PMTU of 1280 is required for IPv6.
b) seems impossible to implement because ICMP messages cannot be reliably mapped to individual packets. Thats why d) isn't an option either (trivial DoS vector). c) is required by the IPv6 RFC, but this is broken. a) is consistent with optional PMTUD, so it seems the sanest choice. -- Florian Weimer <fwei...@bfk.de> BFK edv-consulting GmbH http://www.bfk.de/ Kriegsstraße 100 tel: +49-721-96201-1 D-76133 Karlsruhe fax: +49-721-96201-99 -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------