[OpenAFS-devel] .35 sec rx delay bug?

[Derrick J Brashear]

On Wed, 8 Nov 2006, Rainer Toebbicke wrote:

> Harald Barth wrote:
>> Derrick J Brashear wrote:
>>> Did you have jumbograms enabled? If you disable them, but leave the mtu at 
>>> 9000, are things better? I bet they are.
>> 
>> Derick wrote that things get better when jumbograms are _dis_abled, did he?
>> 
>
> Euh...  I bet he meant with big MTUs. Turning jumbograms off on a standard

Actually, it should cause some degradation always because of how the 
jumbogram gets unrolled on the receive end. It may be overwhelmed by 
issues with window size, but, I'm considering things which can be fixed in 
code efficiency, not in network efficiency (that should also be consider 
but isn't what I am speaking to)

I'd rather just send our best guess at path mtu minus header payloaded 
packets, and not jumbograms.