You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cxf.apache.org by "Malisetti, Ramanjaneyulu" <Ra...@ca.com> on 2012/01/06 13:38:34 UTC
performance issue
Hi,
We have scenario where CXF is just started but there is no activity. But CPU cycles are consumed by Jetty selectors started by CXF endpoints. The difference between with and without CXF on process performance is in order of 300%. We are using CXF 2.3.
Are missing any parameter either at Jetty or CXF? Something seems obviously wrong with what we have.
Did anyone observe this?
Regards
Raman
RE: performance issue
Posted by "Malisetti, Ramanjaneyulu" <Ra...@ca.com>.
Thank you Freeman,
I’ll check at my end.
Regards
Raman
From: Freeman Fang [mailto:freeman.fang@gmail.com]
Sent: Monday, January 16, 2012 11:07 AM
To: users@cxf.apache.org
Cc: Malisetti, Ramanjaneyulu
Subject: Re: performance issue
Hi,
FYI, I've encountered high CPU issue when enable SSL with CXF endpoint and client side use IE to connect this endpoint, this is actually a jetty bug[1], which get fixed in jetty 7.5.1 and afterwards, so definitely try the CXF 2.5.x which already use jetty 7.5.3. IIRC, jetty have fixed several high CPU usage related issue in 7.5.x, so the issue you saw here MAY already get resolved from jetty side.
[1]http://jira.codehaus.org/browse/JETTY-1435
Freeman
On 2012-1-7, at 上午3:20, Daniel Kulp wrote:
On Friday, January 06, 2012 12:38:34 PM Malisetti, Ramanjaneyulu wrote:
Hi,
We have scenario where CXF is just started but there is no
activity. But CPU cycles are consumed by Jetty selectors started by CXF
endpoints. The difference between with and without CXF on process
performance is in order of 300%. We are using CXF 2.3.
Are missing any parameter either at Jetty or CXF? Something seems obviously
wrong with what we have.
Did anyone observe this?
Definitely not something I've ever seen before. Starting up a couple of
the examples that use Jetty and just letting them sit show's 0 CPU usage.
Can you create an example that shows this. Even using 2.3.8 would be fine.
--
Daniel Kulp
dkulp@apache.org - http://dankulp.com/blog
Talend Community Coder - http://coders.talend.com
---------------------------------------------
Freeman Fang
FuseSource
Email:f<ma...@fusesource.com>
Web: fusesource.com<http://fusesource.com/>
Twitter: freemanfang
Blog: http://freemanfang.blogspot.com
Re: performance issue
Posted by Freeman Fang <fr...@gmail.com>.
Hi,
FYI, I've encountered high CPU issue when enable SSL with CXF endpoint
and client side use IE to connect this endpoint, this is actually a
jetty bug[1], which get fixed in jetty 7.5.1 and afterwards, so
definitely try the CXF 2.5.x which already use jetty 7.5.3. IIRC,
jetty have fixed several high CPU usage related issue in 7.5.x, so the
issue you saw here MAY already get resolved from jetty side.
[1]http://jira.codehaus.org/browse/JETTY-1435
Freeman
On 2012-1-7, at 上午3:20, Daniel Kulp wrote:
> On Friday, January 06, 2012 12:38:34 PM Malisetti, Ramanjaneyulu
> wrote:
>> Hi,
>> We have scenario where CXF is just started but there is no
>> activity. But CPU cycles are consumed by Jetty selectors started by
>> CXF
>> endpoints. The difference between with and without CXF on process
>> performance is in order of 300%. We are using CXF 2.3.
>>
>> Are missing any parameter either at Jetty or CXF? Something seems
>> obviously
>> wrong with what we have.
>>
>> Did anyone observe this?
>
> Definitely not something I've ever seen before. Starting up a
> couple of
> the examples that use Jetty and just letting them sit show's 0 CPU
> usage.
>
> Can you create an example that shows this. Even using 2.3.8 would
> be fine.
>
>
> --
> Daniel Kulp
> dkulp@apache.org - http://dankulp.com/blog
> Talend Community Coder - http://coders.talend.com
---------------------------------------------
Freeman Fang
FuseSource
Email:ffang@fusesource.com
Web: fusesource.com
Twitter: freemanfang
Blog: http://freemanfang.blogspot.com
Re: performance issue
Posted by Daniel Kulp <dk...@apache.org>.
On Friday, January 06, 2012 12:38:34 PM Malisetti, Ramanjaneyulu wrote:
> Hi,
> We have scenario where CXF is just started but there is no
> activity. But CPU cycles are consumed by Jetty selectors started by CXF
> endpoints. The difference between with and without CXF on process
> performance is in order of 300%. We are using CXF 2.3.
>
> Are missing any parameter either at Jetty or CXF? Something seems obviously
> wrong with what we have.
>
> Did anyone observe this?
Definitely not something I've ever seen before. Starting up a couple of
the examples that use Jetty and just letting them sit show's 0 CPU usage.
Can you create an example that shows this. Even using 2.3.8 would be fine.
--
Daniel Kulp
dkulp@apache.org - http://dankulp.com/blog
Talend Community Coder - http://coders.talend.com
Re: performance issue
Posted by David Karlsen <da...@gmail.com>.
I would upgrade to a later version (latest 2.4 or latest 2.5) to see
if that helps.
2012/1/6 Malisetti, Ramanjaneyulu <Ra...@ca.com>:
> Hi,
> We have scenario where CXF is just started but there is no activity. But CPU cycles are consumed by Jetty selectors started by CXF endpoints. The difference between with and without CXF on process performance is in order of 300%. We are using CXF 2.3.
>
> Are missing any parameter either at Jetty or CXF? Something seems obviously wrong with what we have.
>
> Did anyone observe this?
>
> Regards
> Raman
--
--
David J. M. Karlsen - http://www.linkedin.com/in/davidkarlsen