Subject: Re: Path MTU discovery
To: None <tech-net@NetBSD.ORG>
From: Michael Richardson <>
List: tech-net
Date: 07/16/1997 02:38:19
>>>>> "Wolfgang" == Wolfgang Rupprecht <> writes:
    Wolfgang> mo@UU.NET (Mike O'Dell) writes:
    >> but if you put an FDDI interface into your box and have it talk
    >> to another server, how will your hard-wired MTU of 1460 take
    >> advantage of half as many packets per NFS transfer??

    Wolfgang> Under NetBSD the MTU is specifiable on a per-routing
    Wolfgang> table entry basis.  If I had an FDDI I'd simply allocate
    Wolfgang> a network for the FDDI and set that MTU to 4k (or
    Wolfgang> whatever the exact number is).  Adding the following to
    Wolfgang> the rc file, sure is faster than adding the
    Wolfgang> MTU-discovery code to the kernel.

    Wolfgang> 	route add -mtu 4096 -net x.x.x.0 myinterface

  Yes, but that isn't the point of PMTU discovery.

  The point is to figure out that there are FDDI networks from A<->B
and so that *NON-LOCAL* packets can be bigger.
  In any case, the reason I wanted PMTU code (which I understand now
that FreeBSD has) is because I want to determine if my implementation
of PMTU ICMP's in an IPsec security gateway works correctly.

] It isn't that sun never sets; rather dawn and dusk are united | one quark   [
]   Michael Richardson, Sandelman Software Works, Ottawa, ON    | two quark   [
] | red q blue q[
] panic("Just another NetBSD/notebook using, kernel hacking, security guy");  [