RE: End System PMTUD behavior question

["Hemant Singh (shemant)" <shemant@cisco.com>]

Jeffrey,

Both IPv4 and IPv6 standards recommend not to fragment.  So the first
impact on the system behavior for host receiving an PTB message is to
try and not fragment.  I have not paid attention to older RFC's.  I am
working off of RFC 4821 that recommends a host keep "conceptual"
variables in cache at the IP layer and all other layers of code on the
host may access this cache to get the current value of the MTU to be
used by host.  See following text from RFC 4821.

[Furthermore, the natural mechanism to 
share Path MTU information between concurrent or subsequent connections 
is a path information cache in the IP layer. The various 
Packetization Protocols need to have the means to access and update 
the shared cache in the IP layer. This memo describes PLPMTUD in 
terms of its primary subsystems without fully describing how they 
are assembled into a complete implementation.]

Also see section 5 of RFC 4821.

What UNH has tested for current host behavior and so have you is because
host are following older PMTU RFCs like 1981.  Also, I agree with you on
RFC 1981 that host may fragment if there is no easy means for the host's
layers to pass MTU information back and forth.  My 2 cents is for new
hosts to follow RFC 4821 and not fragment packets.

Thanks

Hemant