Ouf. Be careful with this. The correspondence between OS-level threads and Java threads changes a lot depending on O.S., JVM, and configuration. See http://java.sun.com/docs/hotspot/threads/threads.html for more info (Solaris-specific, but you'll get the picture).

--
Salut,

Jordi.

Jean-Sebastien Morisset wrote:
On Thu, Oct 02, 2003 at 10:43:10AM +0100, BAZLEY, Sebastian wrote:

Your posting in JMeter user shows that you have included some throughput
controllers - perhaps that is contributing to the ceiling in throughput rate
- but it would not explain why 2 is better than 1.


I just re-ran the 2x20 test with jmeter in non-gui mode. While it was
running, I executed the following command for a few seconds:

while : ; do ps -ef|grep '[0-9] java'|egrep -v 'grep'|wc -l; done > threads.txt

And here's the output:

1 1 1 1 1 1 1 1 1 11 7 1 1 1 27 1 1 1 1 1 1 1 1 1 1 1 1 24 1 1 26 11 28
1 10 1 1 1 1 1 1 1 1 1 1 1 1 27 26 1 1 1 1 1 1 1 1 1 1 1 1 1 1 28 1 1 1
1 1 1 1 13 1 1 1 17 1 1 1 21 1 1 1 1 26 1 18 12 1 1 1 1 1 1 1 1 1 1 1 1
1 24 1 13 1 28 1 1 1 9 1 1 9 1 1 1 1 1 1 1 1 1 1 1 1 1 14 6 1 1 1 1 1 1
1 1 6 1 1 1 25 1 1 1 28 1 20 1 1 1 1 18 1 1 1 1 1 1 10 1 24 1 1 1 1

It looks to me like the threads aren't alive very long, which is
probably why I'm not seeing a high CPU usage.

I'm only using a single thread group, with 20 threads in this example.
Would I be better off spreading the test plan between 2+ thread
groups, and 'running each thread group seperately'? I have a ratio of
hits to respect, so this would complicate things a little bit more. :-)

Thanks,
js.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to