lkml.org 
[lkml]   [1998]   [Aug]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: Strange TCP behavior
From
Date
Bernd Paysan <bernd.paysan@gmx.de> writes:

[...]
> so it seems to be connected with an 128K limit (I tried other values, and
> indeed, once you fill 128k with small chunks, the kernel seems urged to
> wait).
>
> Is this a bug/feature? Has anybody an idea why there is an 128k limit, and
> why it doesn't matter once the chunk size is 1024 bytes or larger? I don't
> think it affects real world benchmarks, but who knows.


Seems to be a buffer management problem. When one changes the default
buffer sizes to 128K (in /proc/sys/net/core/[rw]mem_{max,default}) then
the degradation is at the start, not in the middle. I does not seem to
be prune_queue() related too - prune_queue is never called.
-Andi

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html

\
 
 \ /
  Last update: 2005-03-22 13:43    [from the cache]
©2003-2011 Jasper Spaans. Advertise on this site