On Thu, Apr 19, 2012 at 3:01 AM, Markus Walther <dr.marikos@gmail.com> wrote:
> Hi agentzh,
>
> your advice and pointers regarding weighttp and ab proved most helpful!
>
> I now get 24804 req/s with the same weighttp invocation as yours, and
> 27254.8 req/s with an updated ab invocation!
>
Yay! :D
> I suggest to add the following text to your openresty Benchmark section:
>
> "Note that MacOS X Lion has known issues that need to be fixed on your
> system before attempting to replicate our benchmark results.
>
> In brief, raise the number of available ephemeral ports using <a
> href="http://serverfault.com/questions/145907/does-mac-os-x-throttle-the-rate-of-socket-creation";>this
> fix.</a>
>
> Compile an up-to-date version of ab (Apache's benchmark tool)
> according to <a
> href="http://superuser.com/questions/323840/apache-bench-test-erroron-os-x-apr-socket-recv-connection-reset-by-peer-54";>this</a>.
>
> Then, <code>ab -k -c10 -n10000 -t1 -r 'http://127.0.0.1:8080/'</code>
> will deliver benchmark results.
>
> Or, use an alternative lightweight HTPP load-testing tool <a
> href="http://redmine.lighttpd.net/projects/weighttp/wiki";>weighttp</a>
> and the invocation <code>weighttp -k -c10 -n10000
> 'http://127.0.0.1:8080/'</code> for benchmarking."
>
Already added your note to the Benchmark page:
http://openresty.org/#Benchmark
Thank you very much for your contribution!
Best regards,
-agentzh
