[OpenAFS-devel] nojumbo on server doesn't disable jumbograms on clients

[Jeffrey Hutzelman]

On Saturday, April 15, 2006 02:05:52 PM +0200 Harald Barth 
<haba@pdc.kth.se> wrote:

>
>> While tracing a one-way performance problem with our AFS server (MTU
>> 1500), I  found that on the client side (MTU 8000), every few packets it
>> tried to send a  jumbogram, overriding the negotiated MTU of 1500.
>
> * rx can pack more than one rx message into a datagram.
> * rx can make datagrams > MTU.
> Both these "features" have been called jumbogram. I think
> the second one is not making things better with todays
> computers (fast enough CPUs) and network structure
> (fragmentation and MTU discovery often broken).

And to make matters worse, _neither_ of them is what people outside the AFS 
community mean when they say "jumbogram".  Those people are referring to 
the practice of using "oversized" ethernet packets, such as the original 
poster's interface MTU of 8000 octets.  That practice _can_ be beneficial, 
as long as such oversized packets are used only when the path MTU is 
actually that large.



-- Jeffrey T. Hutzelman (N3NHS) <jhutz+@cmu.edu>
   Sr. Research Systems Programmer
   School of Computer Science - Research Computing Facility
   Carnegie Mellon University - Pittsburgh, PA