On 22/12/13 14:33, Fred Klassen wrote:
>> Some notes:
>>
>> Some old annoyances are still around (
>> https://github.com/synfinatic/tcpreplay/pull/9 ), now trying to figure
>> out how they apply to the new code base and fix them :-). They're quite
>> easy to run into, run tcpreplay
>> with --loop 0 with really small (1-10 packets is typical) pcaps.
> I'll try to look into it. If you find a fix, make sure you put the pull
> request
> on the new root repo https://github/appneta/tcpreplay. It is much easier to
> Handle the pull request.
>
Okie, last two issues seemed to be still relevant (pull request sent) so
my wrapper gets instead of (trying to use --stats 1, starting from
10Mbps and dropping to 0 in 20% increments every 5 secs):
[root@the modules]# node loadgen.js -i virbr0 -r out.pcap -M 10 --time 5
--drop 20
{"packets":74509,"bytes":5969054,"seconds":"4.07","bps":"1249993.20","Mbps":"9.99","pps":"15603.10"}
{"packets":134122,"bytes":10744744,"seconds":"8.14","bps":"999997.90","Mbps":"7.99","pps":"12482.56"}
{"packets":178802,"bytes":14324048,"seconds":"12.21","bps":"749999.40","Mbps":"5.99","pps":"9362.14"}
{"packets":208586,"bytes":16710024,"seconds":"16.28","bps":"499999.60","Mbps":"3.99","pps":"6241.46"}
with patches:
[root@the modules]# node loadgen.js -i virbr0 -r out.pcap -M 10 --time 5
--drop 20
{"packets":15606,"bytes":1250060,"seconds":"1.00","bps":"1249998.70","Mbps":"9.99","pps":"15605.23"}
{"packets":31210,"bytes":2500100,"seconds":"2.00","bps":"1249998.70","Mbps":"9.99","pps":"15604.36"}
{"packets":46813,"bytes":3750102,"seconds":"3.00","bps":"1249999.40","Mbps":"9.99","pps":"15603.90"}
{"packets":62416,"bytes":5000120,"seconds":"4.00","bps":"1249999.30","Mbps":"9.99","pps":"15603.61"}
{"packets":74679,"bytes":5982574,"seconds":"4.07","bps":"1249998.70","Mbps":"9.99","pps":"15603.42"}
{"packets":87164,"bytes":6982596,"seconds":"5.07","bps":"999997.00","Mbps":"7.99","pps":"12484.68"}
{"packets":99646,"bytes":7982604,"seconds":"6.07","bps":"999998.50","Mbps":"7.99","pps":"12483.29"}
{"packets":112129,"bytes":8982618,"seconds":"7.07","bps":"999999.60","Mbps":"7.99","pps":"12483.14"}
{"packets":124611,"bytes":9982630,"seconds":"8.07","bps":"999999.50","Mbps":"7.99","pps":"12482.81"}
{"packets":134264,"bytes":10755852,"seconds":"8.14","bps":"999994.90","Mbps":"7.99","pps":"12482.97"}
...
(which someone might eventually use to plot a nice graph :-) )
After more investigation, looks like no regression in cpu usage after
all. Small files do need --preload-pcap to perform (which -K used to
do), but also 3.4.x did use up lots of CPU at even 1 pps (but I agree,
better be accurate)
Also did a build test on RHEL 6, beta1 worked there
(--enable-dynamic-link, anything but bundled autoopts would be a world
of hurt there :-) )
------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk
_______________________________________________
Tcpreplay-users mailing list
Tcpreplay-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/tcpreplay-users
Support Information: http://tcpreplay.synfin.net/trac/wiki/Support
