RE: iperf 1.7.0 thread bug.
I am having a problem with iperf
iperf version 2.0.0 (23 May 2004) single threaded
On a Sun (Newisys) V20Z AMD Opteron 64 bit machine running Linux 2.6.6
The problem is with the -i (interval option). If I enter -i 5 I get enormous amounts of terminal output with lines like:
2iepm --at-- v20z-1:~>iperf -c 10.10.10.3 -t 5 -i 5
------------------------------------------------------------
Client connecting to 10.10.10.3, TCP port 5001
TCP window size: 128 MByte (default)
------------------------------------------------------------
[ 3] local 10.10.10.1 port 33365 connected with 10.10.10.3 port 5001
[ 3] 0.0--1099623697.1 sec 0.00 Bytes -0.00 bits/sec
[ 3] -1099623697.1--1094623697.1 sec 0.00 Bytes 0.00 bits/sec
[ 3] -1094623697.1--1089623697.1 sec 0.00 Bytes 0.00 bits/sec
[ 3] -1089623697.1--1084623697.1 sec 0.00 Bytes 0.00 bits/sec
[ 3] -1084623697.1--1079623697.1 sec 0.00 Bytes 0.00 bits/sec
[ 3] -1079623697.1--1074623697.1 sec 0.00 Bytes 0.00 bits/sec
It appears specifying the interval in nanoseconds, helps a bit (i.e. gets rid of a lot of junky output)e.g.
5iepm --at-- v20z-1:~>iperf -c 10.10.10.3 -t 20 -i 5000000000
------------------------------------------------------------
Client connecting to 10.10.10.3, TCP port 5001
TCP window size: 128 MByte (default)
------------------------------------------------------------
[ 3] local 10.10.10.1 port 33367 connected with 10.10.10.3 port 5001
[ 3] 0.0--1099624244.8 sec 0.00 Bytes -0.00 bits/sec
[ 3] 0.0-20.0 sec 4.72 GBytes 2.03 Gbits/sec
But note the interval gives 0.00 bits/sc and the seconds are wrong.
-----Original Message-----
From: owner-iperf-users --at-- dast.nlanr.net [mailto:owner-iperf-users --at-- dast.nlanr.net] On Behalf Of Mitch Kutzko
Sent: Wednesday, October 06, 2004 11:27 AM
To: Iperf Users
Subject: iperf 1.7.0 thread bug.
>Subject: iperf 1.7.0 thread bug.
>From: Stephen Hemminger <shemminger --at-- osdl.org>
>To: dast --at-- nlanr.net
>Cc: iperf-users --at-- dast.nlanr.net, netdev --at-- oss.sgi.com
>Content-Type: multipart/mixed; boundary="=-zFU1N+103e8NII/wdZk7"
>Organization: Open Source Development Lab
>Date: Tue, 05 Oct 2004 11:41:18 -0700
>
>The iperf 1.7.0 hangs on AMD64, after several false starts the problem
>turned out to be that the Thread wrapper was incorrectly assumuing that
>pthread wasn't present (in Speaker.cpp), so some data sizes get messed
>up. Bedlam ensuses.
>
>This fixes it.
>
>
>--=-zFU1N+103e8NII/wdZk7
>Content-Description:
>Content-Disposition: attachment; filename=iperf-1.7.0.patch
>Content-Type: text/x-patch; charset=us-ascii
>Content-Transfer-Encoding: 7bit
>
>--- lib/Thread.hpp.orig 2004-10-05 11:33:14.638098735 -0700
>+++ lib/Thread.hpp 2004-10-05 11:34:30.160583668 -0700
> --at-- --at-- -56,7 +56,9 --at-- --at--
> #ifndef THREAD_H
> #define THREAD_H
>
>-
>+#ifdef HAVE_CONFIG_H
>+ #include "config.h"
>+#endif
>
> #if defined( HAVE_POSIX_THREAD )
>
>
>--=-zFU1N+103e8NII/wdZk7--
>
>
--
Mitch Kutzko | mitch --at-- dast.nlanr.net | mitch --at-- ncsa.uiuc.edu | 217-333-1199
Project: http://dast.nlanr.net/ | Personal: http://hobbes.ncsa.uiuc.edu/