RE: Question about iperf -B option
I wanted to achieve the same thing, so I patched the standard 2.6.24
kernel from kernel.org with the send to self patch and a real time
patch. The real time patch I need in order to get higher accuracy in
round trip time measurements. Anyhow, it doesn't work. The loopback
interface is still used, as reported by tcpdump -i lo while running
iperf with the -B option.
Is there any simple way to troubleshoot this? I don't really have a clue
what to do next.
/ Daniel
On Wed, 2008-01-30 at 11:29 +0100, David Ruiz wrote:
> Thanks a lot. I patched the kernel and it works now.
>
> David
>
> -----Original Message-----
> From: Stephen Hemminger [mailto:shemminger --at-- linux-foundation.org]
> Sent: miÃrcoles, 30 de enero de 2008 4:17
> To: iperf-users --at-- dast.nlanr.net
> Cc: david.ruiz --at-- gigle.biz
> Subject: Re: Question about iperf -B option
>
> On Tue, 29 Jan 2008 23:16:59 +0100 (CET)
> "David Ruiz" <david.ruiz --at-- gigle.biz> wrote:
>
> > Hello,
> >
> > I am trying to do some performance testing with two external powerline
> > adaptors. For this, I have a single PC with Linux, and two ethernet cards,
> > each one connected to each of the adaptors. My objective is to send data
> > generated with iperf through eth1, so that it goes to the adaptor 1 and
> > via powerline arrives to adaptor 2 and then to eth2.
> >
> > In order to do this i configure the server as :
> > iperf -s -u -B 10.10.1.2 (10.10.1.2 is the IP of eth2)
> >
> > and the client as :
> > iperf -c 10.10.1.2 -u -B 10.10.1.1 (10.10.1.1 is the IP of eth1)
> >
> > As far as I can tell, the data arrives to the server, but they use the
> > loopback. Does anybody know if this can be done (and how??)
> >
> > Thank you
> >
> >
> >
>
> there is a non-standard kernel patch called "send to self".
> you can find it with some searching.
>