You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jetspeed-user@portals.apache.org by ramaj <ra...@gmail.com> on 2007/11/02 10:39:25 UTC
RE: RAM size doesnot decrease....
hello scott,
Herer is procedure for stress testing by JMETER tool.
JMETER TOOL
Requirement:
1-jdk1.4
2.jakarta-jmeter-2.3RC3.zip(for windows)
3.jakarta-jmeter-2.3RC3.tgz(for linux
To Start on Window:
1.Double click on ApacheJmeter.jar(for windows)
Procedure:
1.Right click on Test Plan click ->Thread Group
--adjust the thread group and loop count as u wish....Ramp up period must
be 0.
2.Rightclick on Thread Group->Click add->Logic Controller->Loop Controller
3.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP cookie
manager
--check clear cookies each iteration
4.Rightclick on LoopControoler-> Click->add->ConfigElement->CSV Data Set
Config
--Configure the CSV Data Source
Filename : filename.txt
File encoding :
VaraibleNames : USER,PASS
Delimiter : ,
5.Rightclick on LoopControoler-> Click->add->ConfigElement->User Defined
Variables
--User Defined Variables
Name Value
USER PASS
6.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP Request
Default
--IP
--port number (default 80)
--path /webportal
7.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
--Method POST
--path-/webportal/portal/login/proxy
--add parameters with request
Name value
org.apache.jetspeed.username ${USER}
org.apache.jetspeed.password ${PASS}
8.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
--Method POST
--path-/webportal/login/j_security_check
--add parameters with request
Name value
j_username ${USER}
j_password ${PASS}
9.Create the filename.txt with user and password
e.g.
ram,password,
scott,passowrd,
...
This file put in "bin" directory of JMETER.
10.Rightclick on LoopControoler-> Click->add->Listeners->as u wish
(e.g.aggregate results).
11.Menu->Run->start.
For Linux:
To start:
run by "jmeter" on command prompt
Procedure
Same as above.
Weaver, Scott-2 wrote:
>
> Hi Ramj,
>
> This is a known issue. The major spots for this leakage are
> org.apache.jetspeed.profiler.impl.JetspeedProfilerImpl and
> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl.
> The cause for the leakage in both of these cases is the fact that both
> objects are caching large amounts of information in a standard hashmaps,
> which is really killing memory. I am working on replacing these with
> true cache implementations. I am getting ready to test first one in the
> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl in a
> running portal.
>
> Do you happen to have any profiler dumps of your JVM at or near the time
> of the OOM, do you? I have my own and one other from Ethan Adams, both
> of which point to the same two spots. I want to make sure that we get
> all possible leaks ironed out asap. Also, could you describe your
> JMeter profile/load plan?
>
> Thanks,
> -scott
>
>
>> -----Original Message-----
>> From: ramaj [mailto:ramjmeter@gmail.com]
>> Sent: Friday, October 26, 2007 5:15 AM
>> To: jetspeed-user@portals.apache.org
>> Subject: RAM size doesnot decrease....
>>
>>
>> Hi..
>> I have installed Jetspeed2.1.
>> I am doing stress testing using JMETER.
>>
>> H/W:
>> Linux m/c with 8 GB RAM.
>>
>> userd RAM size increases as requests increases...(top command on
> linux)
>> 1.For 1000 requests:
>> Mem: 8309184k total, 3231536k used, 5077648k free,
>> 371160k buffers
>>
>> 2.For 2000 requests:
>> Mem: 8309184k total, 3710064k used, 4599120k free,
>> 371660k buffers
>> 3.For 3000 requests:
>> Mem: 8309184k total, 4190320k used, 4118864k free,
>> 371816k buffers
>> 4.For 5000 requests:
>> Mem: 8309184k total, 4923192k used, 3385992k free,
>> 372084k buffers
>> 5.For 7000 requests:
>> Mem: 8309184k total, 4985592k used, 3323592k free,
>> 372260k buffers
>> 6.For 9000 requests:
>> Mem: 8309184k total, 5037640k used, 3271544k free,
>> 372616k buffers
>>
>> For 9000 it gives outofmemory:java heap size
>>
>>
>> But after 1000 requests or 5000 requests threads should be free..OR
>> objects
>> have to be collected by GC.
>> But here this is not happen..........
>>
>> OR..
>> here Object not collected by GC.............?
>>
>>
>> plz tell me.............help me..
>>
>> what is the reason.?
>> what is the solutions..
>>
>>
>> Thanks RAMJ
>>
>>
>>
>> --
>> View this message in context: http://www.nabble.com/RAM-size-doesnot-
>> decrease....-tf4696063.html#a13423481
>> Sent from the Jetspeed - User mailing list archive at Nabble.com.
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
>> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
>
>
>
--
View this message in context: http://www.nabble.com/RAM-size-doesnot-decrease....-tf4696063.html#a13544519
Sent from the Jetspeed - User mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
For additional commands, e-mail: jetspeed-user-help@portals.apache.org
RE: RAM size doesnot decrease....
Posted by "Weaver, Scott" <we...@ugs.com>.
I was going to ask this same question as I think the memory leak only happens when there are many different profiles hitting the portal.
-scott
> -----Original Message-----
> From: Joachim Müller [mailto:joachim@wemove.com]
> Sent: Friday, November 02, 2007 8:24 AM
> To: Jetspeed Users List
> Subject: Re: RAM size doesnot decrease....
>
> Hi ramaj,
>
> can you say something about the number of users you have in your
> filename.txt?
>
> Regards,
> Joachim
>
> ramaj schrieb:
> > hello scott,
> >
> > Herer is procedure for stress testing by JMETER tool.
> >
> > JMETER TOOL
> >
> > Requirement:
> > 1-jdk1.4
> > 2.jakarta-jmeter-2.3RC3.zip(for windows)
> > 3.jakarta-jmeter-2.3RC3.tgz(for linux
> >
> > To Start on Window:
> >
> > 1.Double click on ApacheJmeter.jar(for windows)
> >
> >
> > Procedure:
> >
> > 1.Right click on Test Plan click ->Thread Group
> > --adjust the thread group and loop count as u wish....Ramp up period
> must
> > be 0.
> > 2.Rightclick on Thread Group->Click add->Logic Controller->Loop
> Controller
> > 3.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP cookie
> > manager
> > --check clear cookies each iteration
> > 4.Rightclick on LoopControoler-> Click->add->ConfigElement->CSV Data Set
> > Config
> > --Configure the CSV Data Source
> > Filename : filename.txt
> > File encoding :
> > VaraibleNames : USER,PASS
> > Delimiter : ,
> > 5.Rightclick on LoopControoler-> Click->add->ConfigElement->User Defined
> > Variables
> > --User Defined Variables
> > Name Value
> > USER PASS
> >
> > 6.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP Request
> > Default
> > --IP
> > --port number (default 80)
> > --path /webportal
> > 7.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
> > --Method POST
> > --path-/webportal/portal/login/proxy
> > --add parameters with request
> > Name value
> > org.apache.jetspeed.username ${USER}
> > org.apache.jetspeed.password ${PASS}
> > 8.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
> > --Method POST
> > --path-/webportal/login/j_security_check
> > --add parameters with request
> > Name value
> > j_username ${USER}
> > j_password ${PASS}
> > 9.Create the filename.txt with user and password
> > e.g.
> > ram,password,
> > scott,passowrd,
> > ...
> > This file put in "bin" directory of JMETER.
> >
> > 10.Rightclick on LoopControoler-> Click->add->Listeners->as u wish
> > (e.g.aggregate results).
> > 11.Menu->Run->start.
> >
> > For Linux:
> > To start:
> > run by "jmeter" on command prompt
> > Procedure
> > Same as above.
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Weaver, Scott-2 wrote:
> >> Hi Ramj,
> >>
> >> This is a known issue. The major spots for this leakage are
> >> org.apache.jetspeed.profiler.impl.JetspeedProfilerImpl and
> >> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl.
> >> The cause for the leakage in both of these cases is the fact that both
> >> objects are caching large amounts of information in a standard
> hashmaps,
> >> which is really killing memory. I am working on replacing these with
> >> true cache implementations. I am getting ready to test first one in
> the
> >> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl in
> a
> >> running portal.
> >>
> >> Do you happen to have any profiler dumps of your JVM at or near the
> time
> >> of the OOM, do you? I have my own and one other from Ethan Adams, both
> >> of which point to the same two spots. I want to make sure that we get
> >> all possible leaks ironed out asap. Also, could you describe your
> >> JMeter profile/load plan?
> >>
> >> Thanks,
> >> -scott
> >>
> >>
> >>> -----Original Message-----
> >>> From: ramaj [mailto:ramjmeter@gmail.com]
> >>> Sent: Friday, October 26, 2007 5:15 AM
> >>> To: jetspeed-user@portals.apache.org
> >>> Subject: RAM size doesnot decrease....
> >>>
> >>>
> >>> Hi..
> >>> I have installed Jetspeed2.1.
> >>> I am doing stress testing using JMETER.
> >>>
> >>> H/W:
> >>> Linux m/c with 8 GB RAM.
> >>>
> >>> userd RAM size increases as requests increases...(top command on
> >> linux)
> >>> 1.For 1000 requests:
> >>> Mem: 8309184k total, 3231536k used, 5077648k free,
> >>> 371160k buffers
> >>>
> >>> 2.For 2000 requests:
> >>> Mem: 8309184k total, 3710064k used, 4599120k free,
> >>> 371660k buffers
> >>> 3.For 3000 requests:
> >>> Mem: 8309184k total, 4190320k used, 4118864k free,
> >>> 371816k buffers
> >>> 4.For 5000 requests:
> >>> Mem: 8309184k total, 4923192k used, 3385992k free,
> >>> 372084k buffers
> >>> 5.For 7000 requests:
> >>> Mem: 8309184k total, 4985592k used, 3323592k free,
> >>> 372260k buffers
> >>> 6.For 9000 requests:
> >>> Mem: 8309184k total, 5037640k used, 3271544k free,
> >>> 372616k buffers
> >>>
> >>> For 9000 it gives outofmemory:java heap size
> >>>
> >>>
> >>> But after 1000 requests or 5000 requests threads should be free..OR
> >>> objects
> >>> have to be collected by GC.
> >>> But here this is not happen..........
> >>>
> >>> OR..
> >>> here Object not collected by GC.............?
> >>>
> >>>
> >>> plz tell me.............help me..
> >>>
> >>> what is the reason.?
> >>> what is the solutions..
> >>>
> >>>
> >>> Thanks RAMJ
> >>>
> >>>
> >>>
> >>> --
> >>> View this message in context: http://www.nabble.com/RAM-size-doesnot-
> >>> decrease....-tf4696063.html#a13423481
> >>> Sent from the Jetspeed - User mailing list archive at Nabble.com.
> >>>
> >>>
> >>> ---------------------------------------------------------------------
> >>> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
> >>> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
> >>
> >> ---------------------------------------------------------------------
> >> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
> >> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
> >>
> >>
> >>
> >
>
>
> --
>
> . frankfurt am main, 10°c, zur zeit aufgelockert
> bewölkt in 244 m. bewölkt 640 m. die sichtweite
> reicht 6.0 km.
>
> < joachim müller
> joachim@wemove.com
> t +49 69 759003 11
>
> wemove digital solutions gmbh
> eschersheimer landstr. 5-7
> 60322 frankfurt am main
>
> amtsgericht frankfurt am main, hrb 53992
> geschäftsführer joachim müller, stefan hartmann
>
> wemove digital solutions
> www.wemove.com
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
For additional commands, e-mail: jetspeed-user-help@portals.apache.org
Re: RAM size doesnot decrease....
Posted by Joachim Müller <jo...@wemove.com>.
Hi ramaj,
can you say something about the number of users you have in your
filename.txt?
Regards,
Joachim
ramaj schrieb:
> hello scott,
>
> Herer is procedure for stress testing by JMETER tool.
>
> JMETER TOOL
>
> Requirement:
> 1-jdk1.4
> 2.jakarta-jmeter-2.3RC3.zip(for windows)
> 3.jakarta-jmeter-2.3RC3.tgz(for linux
>
> To Start on Window:
>
> 1.Double click on ApacheJmeter.jar(for windows)
>
>
> Procedure:
>
> 1.Right click on Test Plan click ->Thread Group
> --adjust the thread group and loop count as u wish....Ramp up period must
> be 0.
> 2.Rightclick on Thread Group->Click add->Logic Controller->Loop Controller
> 3.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP cookie
> manager
> --check clear cookies each iteration
> 4.Rightclick on LoopControoler-> Click->add->ConfigElement->CSV Data Set
> Config
> --Configure the CSV Data Source
> Filename : filename.txt
> File encoding :
> VaraibleNames : USER,PASS
> Delimiter : ,
> 5.Rightclick on LoopControoler-> Click->add->ConfigElement->User Defined
> Variables
> --User Defined Variables
> Name Value
> USER PASS
>
> 6.Rightclick on LoopControoler-> Click->add->ConfigElement->HTTP Request
> Default
> --IP
> --port number (default 80)
> --path /webportal
> 7.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
> --Method POST
> --path-/webportal/portal/login/proxy
> --add parameters with request
> Name value
> org.apache.jetspeed.username ${USER}
> org.apache.jetspeed.password ${PASS}
> 8.Rightclick on LoopControoler-> Click->add->Sampler->HTTP Request
> --Method POST
> --path-/webportal/login/j_security_check
> --add parameters with request
> Name value
> j_username ${USER}
> j_password ${PASS}
> 9.Create the filename.txt with user and password
> e.g.
> ram,password,
> scott,passowrd,
> ...
> This file put in "bin" directory of JMETER.
>
> 10.Rightclick on LoopControoler-> Click->add->Listeners->as u wish
> (e.g.aggregate results).
> 11.Menu->Run->start.
>
> For Linux:
> To start:
> run by "jmeter" on command prompt
> Procedure
> Same as above.
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> Weaver, Scott-2 wrote:
>> Hi Ramj,
>>
>> This is a known issue. The major spots for this leakage are
>> org.apache.jetspeed.profiler.impl.JetspeedProfilerImpl and
>> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl.
>> The cause for the leakage in both of these cases is the fact that both
>> objects are caching large amounts of information in a standard hashmaps,
>> which is really killing memory. I am working on replacing these with
>> true cache implementations. I am getting ready to test first one in the
>> org.apache.jetspeed.container.window.impl.PortletWindowAccessorImpl in a
>> running portal.
>>
>> Do you happen to have any profiler dumps of your JVM at or near the time
>> of the OOM, do you? I have my own and one other from Ethan Adams, both
>> of which point to the same two spots. I want to make sure that we get
>> all possible leaks ironed out asap. Also, could you describe your
>> JMeter profile/load plan?
>>
>> Thanks,
>> -scott
>>
>>
>>> -----Original Message-----
>>> From: ramaj [mailto:ramjmeter@gmail.com]
>>> Sent: Friday, October 26, 2007 5:15 AM
>>> To: jetspeed-user@portals.apache.org
>>> Subject: RAM size doesnot decrease....
>>>
>>>
>>> Hi..
>>> I have installed Jetspeed2.1.
>>> I am doing stress testing using JMETER.
>>>
>>> H/W:
>>> Linux m/c with 8 GB RAM.
>>>
>>> userd RAM size increases as requests increases...(top command on
>> linux)
>>> 1.For 1000 requests:
>>> Mem: 8309184k total, 3231536k used, 5077648k free,
>>> 371160k buffers
>>>
>>> 2.For 2000 requests:
>>> Mem: 8309184k total, 3710064k used, 4599120k free,
>>> 371660k buffers
>>> 3.For 3000 requests:
>>> Mem: 8309184k total, 4190320k used, 4118864k free,
>>> 371816k buffers
>>> 4.For 5000 requests:
>>> Mem: 8309184k total, 4923192k used, 3385992k free,
>>> 372084k buffers
>>> 5.For 7000 requests:
>>> Mem: 8309184k total, 4985592k used, 3323592k free,
>>> 372260k buffers
>>> 6.For 9000 requests:
>>> Mem: 8309184k total, 5037640k used, 3271544k free,
>>> 372616k buffers
>>>
>>> For 9000 it gives outofmemory:java heap size
>>>
>>>
>>> But after 1000 requests or 5000 requests threads should be free..OR
>>> objects
>>> have to be collected by GC.
>>> But here this is not happen..........
>>>
>>> OR..
>>> here Object not collected by GC.............?
>>>
>>>
>>> plz tell me.............help me..
>>>
>>> what is the reason.?
>>> what is the solutions..
>>>
>>>
>>> Thanks RAMJ
>>>
>>>
>>>
>>> --
>>> View this message in context: http://www.nabble.com/RAM-size-doesnot-
>>> decrease....-tf4696063.html#a13423481
>>> Sent from the Jetspeed - User mailing list archive at Nabble.com.
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
>>> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
>> For additional commands, e-mail: jetspeed-user-help@portals.apache.org
>>
>>
>>
>
--
. frankfurt am main, 10°c, zur zeit aufgelockert
bewölkt in 244 m. bewölkt 640 m. die sichtweite
reicht 6.0 km.
< joachim müller
joachim@wemove.com
t +49 69 759003 11
wemove digital solutions gmbh
eschersheimer landstr. 5-7
60322 frankfurt am main
amtsgericht frankfurt am main, hrb 53992
geschäftsführer joachim müller, stefan hartmann
wemove digital solutions
www.wemove.com
---------------------------------------------------------------------
To unsubscribe, e-mail: jetspeed-user-unsubscribe@portals.apache.org
For additional commands, e-mail: jetspeed-user-help@portals.apache.org