jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Philippe Mouawad <philippe.moua...@gmail.com>
Subject Re: Distributed testing - Memory usage
Date Thu, 29 Sep 2011 13:17:01 GMT
Hi Again,
And please confirm (just to have it written :-) ) you were using HTTPClient
4 new implementation.

Regards
Philippe

On Thu, Sep 29, 2011 at 2:57 PM, Philippe Mouawad <
philippe.mouawad@gmail.com> wrote:

> Hello,
> You OOM is related to bug in HTTP HC4 implementation that is fixed in
> 2.5.1-RC3.
>
> So using nightly build will fix your issue.
> But please confirm it does.
>
> Regards
> Philippe
>
>
> On Thu, Sep 29, 2011 at 2:46 PM, Philippe Mouawad <
> philippe.mouawad@gmail.com> wrote:
>
>> Hi again,
>> Thanks for your post I will review it as soon as I have time.
>> Can you run your test with last trunk version ? See nightly builds
>> 2.5.1-RC3
>>
>> From my tests, I don't reproduce your issue with what you specified on
>> Trunk version:
>>
>>    - 5 Threads
>>    - 10000 iterations
>>    - 15 HTTPClient going to jakarata site
>>    - Constant Timer
>>    - -l option
>>
>> So either we fixed it either there is something more in your JMX.
>>
>> Regards
>> Philippe
>>
>>
>> On Thu, Sep 29, 2011 at 2:36 PM, poox <rb@bancha.net> wrote:
>>
>>> Hi Philippe, thanks for your response.
>>>
>>> I already had the HeapDump setup as requested, here's a link to one of
>>> the
>>> many I have from testing:
>>>
>>> http://www.mediafire.com/?r4xqa08chm6y5cq
>>>
>>> With regard to your questions:
>>>
>>> - I have tried differnt numbers of threads from 1 - 20 but in all cases
>>> the
>>> memory usage escalates and eventually oom kills jmeter. Naturally this
>>> takes
>>> a little bit longer with 1 thread as the memory usage is lower and the
>>> ramp
>>> up slightly slower.
>>>
>>> - I can't really share my JMX file as it has private information in it, I
>>> have tried stripping it all back to a single HTTP request however and it
>>> still causes the memory to slowly escalate on the slave servers.
>>>
>>> - I have changed the Xmx value to as high as possible without impacting
>>> on
>>> system performance. At no point am I hitting swap on any of the systems
>>> involved.
>>>
>>> The servers in question are Intel C2D 2.66 CPUs with 2GB RAM, I am using
>>> the
>>> following Xmx value on all servers:
>>> HEAP="-Xms1024m -Xmx1024m"
>>>
>>> with per-sample settings of:
>>> NEW="-XX:NewSize=128m -XX:MaxNewSize=256m"
>>>
>>>
>>>
>>>
>>>
>>>
>>> --
>>> View this message in context:
>>> http://jmeter.512774.n5.nabble.com/Distributed-testing-Memory-usage-tp4849000p4852905.html
>>> Sent from the JMeter - User mailing list archive at Nabble.com.
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
>>> For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>>>
>>>
>>
>>
>> --
>> Cordialement.
>> Philippe Mouawad.
>>
>>
>>
>>
>
>
> --
> Cordialement.
> Philippe Mouawad.
>
>
>
>


-- 
Cordialement.
Philippe Mouawad.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message