Re: network traffic doesn't balence well on recent kernels

Jeff Millar (jeff@wa1hco.mv.com)
Sun, 6 Jun 1999 21:52:31 -0400

I'm generally familiar with TCP/IP's algorithms. The reason I'm asking has
more to do with the fact that Windows 98 and older Linux kernels *work
better*. I find myself returning to W98 to surf and download because it
makes better use of my time...that's not a situation I like...it's just the
way it works now....and maybe reporting it will help someone.

Can anyone think of a way that the kernels changed in the last few months
that would make "link hogging" worse? In the meantime, I'll try some of
Alan's suggestions and report any changes.

jeff

----- Original Message -----
From: Alan Cox <alan@lxorguk.ukuu.org.uk>
To: ron flory <rjflory@feist.com>
Cc: <linux-kernel@vger.rutgers.edu>
Sent: Saturday, June 05, 1999 7:55 PM
Subject: Re: network traffic doesn't balence well on recent kernels

> > Jeff Millar wrote:
> > > When using the internet with late v2.2 and recent v2.3 kernels, the
> > > first process to get the link seems to hog the bandwidth.
> >
> > I've noticed this since 2.0.1x, it might be 'normal' yet unexpected
> > behavior.
>
> Link capture is a normal TCP behaviour unfortunately. You can reduce it a
bit
> on modem links by lowering the tcp window (man route) and/or the driver
> queue length (ifconfig ppp0 txqlen 3 I believe for newer ifconfig).
>
> What actually occurs is that one session starts to drop packets more than
the
> other, it backs off so the other takes up the rest of the bandwidth,
probability
> then says the backed off session will tend to fail to get packets through
> on retry too
>
>
> -
> 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.tux.org/lkml/
>

-
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.tux.org/lkml/