<table cellspacing="0" cellpadding="0" border="0" ><tr><td valign="top" style="font: inherit;">thanks again!<br><br>I'm unable to make use of any of the command options mentioned below.<br>However, i did an md5sum on both netperf binaries and they are a match so i'm hoping that is an adequate confirmation.<br><br>I have made some amendments to the x25 routing statements and this time, i can see a TCP session established between client and server but the error i'm getting back from the client is:<br><br>test1:~$ netperf -H 192.168.1.13 -p 4000<br>TCP STREAM TEST to 192.168.1.13<br>netperf: send_tcp_stream: data socket connect failed: Connection refused<br>&nbsp;port: 32791<br><br>does this relate to the fact that the linux host do not allow receiving data on this ports.<br>There are no firewalls between the two devices.<br><br>--- On <b>Thu, 4/16/09, Rick Jones <i>&lt;rick.jones2@hp.com&gt;</i></b> wrote:<br><blockquote style="border-left: 2px solid
 rgb(16, 16, 255); margin-left: 5px; padding-left: 5px;">From: Rick Jones &lt;rick.jones2@hp.com&gt;<br>Subject: Re: [netperf-talk] Packet size<br>To: echelon360@yahoo.com<br>Cc: netperf-talk@netperf.org<br>Date: Thursday, April 16, 2009, 3:57 PM<br><br><pre>Clive Barker wrote:<br>&gt; <br>&gt; Hi everyone,<br>&gt; <br>&gt; I've increased the debugging on this to validate why i'm still<br>hitting the same issue. Below is the capture. I guess the next obvious thing was<br>to check the version of netperf running on both machines.Is there a command<br>string i can use to extract this? I'm pretty sure both are 2.4.4<br><br>If you are running a new-enough netperf/netserver you can run:<br><br>netperf -V<br><br>and<br><br>netserver -V<br><br>otherwise, since I am an old-time mneomic memory circuits from stone knives and<br>bearskins kind of guy, both binaries should have "what strings" in<br>them so if you have a what command you can do:<br><br>what
 netperf<br>what netserver<br><br>if you don't have a what command, I think you can simulated it with a<br>combination of strings and grep:<br><br>strings netperf | fgrep "@(#)" - take the highest version you can see<br>- don't be surprised if netperf -V and the what strings are not in complete<br>agreement though...<br><br>rick jones<br><br><br>&gt; <br>&gt; <br>&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -d -d -d<br>&gt; Program name: netperf<br>&gt; Local send alignment: 8<br>&gt; Local recv alignment: 8<br>&gt; Remote send alignment: 8<br>&gt; Remote recv alignment: 8<br>&gt; Report local CPU 0<br>&gt; Report remote CPU 0<br>&gt; Verbosity: 1<br>&gt; Debug: 3<br>&gt; Port: 4000<br>&gt; Test name: TCP_STREAM<br>&gt; Test bytes: 0 Test time: 10 Test trans: 0<br>&gt; Host name: 192.168.1.13<br>&gt; <br>&gt; installing catcher for all signals<br>&gt; remotehost is 192.168.1.13 and port 4000<br>&gt; establish_control: entered with 192.168.1.13 and
 4000<br>&gt; resolved the destination...<br>&gt; creating a socket<br>&gt; about to connect<br>&gt; establish_control: connect completes<br>&gt; entered send_request...contents before htonl:<br>&gt; request contents:<br>&gt; 0:             1        0        0        0     |   | |    | |    | |    |<br>&gt; 4:             0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 8:             0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 12:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 16:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 20:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 24:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 28:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 32:            0        0        0        0     | 
   | |    | |    | |   <br>|<br>&gt; 36:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 40:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 44:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 48:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 52:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 56:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 60:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; send_request...contents after htonl:<br>&gt; request contents:<br>&gt; 0:       1000000        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 4:             0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 8:             0        0        0        0     |    | |    | |    | |  
 <br>|<br>&gt; 12:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 16:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 20:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 24:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 28:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 32:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 36:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 40:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 44:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 48:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 52:            0        0        0        0     |    | |    | |    | |  
 <br>|<br>&gt; 56:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; 60:            0        0        0        0     |    | |    | |    | |   <br>|<br>&gt; <br>&gt; send_request: about to send 256 bytes from 0x805f700<br>&gt; recv_response: received a 74 byte response<br>&gt; recv_response: partial response received: 74 bytes<br>&gt; response contents<br>&gt; 0:      fffb01ff fb03fffd 18fffd1f 54727969     &gt;ÿûÿ&lt;<br>&gt;ýÿû&lt; &gt;ýÿ&lt; &gt;iyrT&lt;<br>&gt; 4:      6e672030 35303532 33363136 38303032     &gt;0 gn&lt; &gt;2505&lt;<br>&gt;6163&lt; &gt;2008&lt;<br>&gt; ...&lt; &gt;C %e2e2e0d  a252043 6f6e6e65 6374696f     &gt;<br>&gt; &lt; &gt;enno&lt; &gt;oitc&lt;<br>&gt; 12:     6e207265 66757365 64206279 2072656d     &gt;er n&lt; &gt;esuf&lt;<br>&gt;yb d&lt; &gt;mer &lt;<br>&gt; 16:     6f746520 686f7374  d0a0000        0     &gt; eto&lt; &gt;tsoh&lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 20:            0   
     0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 24:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 28:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 32:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 36:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 40:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 44:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 48:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 52:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;   
 &lt;<br>&gt; 56:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; 60:            0        0        0        0     &gt;    &lt; &gt;    &lt;<br>&gt;    &lt; &gt;    &lt;<br>&gt; <br>&gt; <br>&gt; --- On *Tue, 4/14/09, Rick Jones /&lt;rick.jones2@hp.com&gt;/* wrote:<br>&gt; <br>&gt;     From: Rick Jones &lt;rick.jones2@hp.com&gt;<br>&gt;     Subject: Re: [netperf-talk] Packet size<br>&gt;     To: echelon360@yahoo.com<br>&gt;     Cc: netperf-talk@netperf.org<br>&gt;     Date: Tuesday, April 14, 2009, 11:23 AM<br>&gt; <br>&gt; Clive Barker wrote:<br>&gt;&gt; hi everyone, i am hitting another bump and this time, although i'm<br>&gt; able to get something across, i continue to get the same error message<br>while<br>&gt; sending TCP streams across<br>&gt;&gt; <br>&gt;&gt; any ideas please?<br>&gt;&gt; <br>&gt;&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -t TCP_STREAM -- -m 128 -D <br>&gt;
 recv_response: partial response received: 71 bytes<br>&gt;&gt; <br>&gt;&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -t TCP_STREAM -- -m 512 -D<br>&gt;&gt; recv_response: partial response received: 71 bytes<br>&gt;&gt; <br>&gt;&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -t TCP_STREAM -- -m 10000 -D<br>&gt;&gt; recv_response: partial response<br>&gt;  received: 71 bytes<br>&gt;&gt; <br>&gt;&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -t TCP_STREAM -- -m 10000000<br>-D<br>&gt;&gt; recv_response: partial response received: 71 bytes<br>&gt;&gt; <br>&gt;&gt; test1:~$ netperf -H 192.168.1.13 -p 4000 -t TCP_STREAM -- -m 10000000<br>-D<br>&gt;&gt; recv_response: partial response received: 71 bytes<br>&gt; <br>&gt; Easy bit first - make certain you have the very same version of netperf<br>and<br>&gt; netserver running on both sides.<br>&gt; <br>&gt; Also, if the reason for the non-standard control connection port number is<br>&gt; firewalls, you will also
 have to open a hole or holes for the data<br>connections,<br>&gt; and use test-specific options to constrain them to specific port numbers<br>(ones<br>&gt; other than that used for the control connection).<br>&gt; <br>&gt; Finally, once the send size (-m) is greater than the MSS for the<br>connection,<br>&gt; the -D *should* be a noop.  If you see performance differences for large<br>sends<br>&gt; with and without -D then something is arguably broken in the<br>&gt;  stack WRT its<br>&gt; implementation of the Nagle Algorithm.<br>&gt; <br>&gt; Now, if the versions match, and firewalls aren't an issue, it suggests<br>that<br>&gt; perhaps netserver is terminating for some reason.  You could enable<br>further<br>&gt; debugging with -d on the netperf command line and look in /tmp or /var/tmp<br>for<br>&gt; netserver log files which may include information about what is wrong. <br>You<br>&gt; might also want to start netserver with a -d option for
 additonal<br>debugging.<br>&gt; <br>&gt; happy benchmarking,<br>&gt; <br>&gt; rick jones<br>&gt; <br>&gt; <br><br></pre></blockquote></td></tr></table><br>