Re: [naviserver-devel] [naviserver-commits] naviserver/nsd driver.c, 1.97, 1.98

[Stephen D. <sd...@gm...>]

On Sun, Apr 27, 2008 at 9:39 AM, Vasiljevic Zoran <zv...@ar...> wrote:
>
> On 26.04.2008, at 23:46, Vlad Seryakov wrote:
>
>> I think it was working perfectly fine, new buffer size is set on
>> accepted socket, new one, not the one we listened.
>>
>>
>> I used to play with these options for one of the modules i wrote and
>> at
>> least on Linux it was working fine.
>
> OK. I will back this off. I was unable to get that working
> on my platforms. So I thought to comment this and move
> close to the place where it *might* work.
>
> Good to know that this works on Linux at least...


These options can probably just be dropped:

    http://www.psc.edu/networking/projects/tcptune/#options


For Linux it says:

NB: Recent versions of Linux (version 2.6.17 and later) have full
autotuning with 4 MB maximum buffer sizes. Except in some rare cases,
manual tuning is unlikely to substantially improve the performance of
these kernels over most network paths, and is not generally
recommended


This wasn't the web page I was looking for. There's one somewhere that
describes a project that looked at autotuning TCP buffer sizes and I
remember it said that all major OS' were fixed a couple of years ago,
as a result.

I think this is not needed anymore, and probably just makes things
worse if used.

One less knob!