You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@jmeter.apache.org by wvaibhav <va...@sigmainfo.net> on 2008/04/27 09:04:50 UTC
10k Hits for web site
Hello All,
I am simulating 10K hits using jmeter distrubuted environment (1 master
three slaves) I have planned the task and tried to execute, but having some
trouble in execution... and also wanted to confirm the scenario which i have
executed....and will be improving gradually
I have recorded a script using jmeter proxy. I wanted a rampup period for
the execution to be 10 threads per two seconds. So, I selected 50 threads,
ramp up period 10 and used a loop count for 2000. it takes around 6-7 hours
to complete the job.... If I increase the number of threads and rampup
period and reduce the loop count, i receive out of memory error and server
in distributed environment gets preety slow completion of @ 60% of job say
100 requests in one hour. PF usage is 1.2 GB for 1 GB RAM and cpu usage is
around 55 %. How is the scenario?
How much rampup period jmeter consider after completion of first loop?
whether it considers same rampup period for all the loops?
What are standard maximum number of threads, rampup period and loop count
for heavy loads....
Another question is while recording scripts through jmeter, it collects all
the image files and other requests from the web page to whom we are having
hits. For actual testing whether we are suppose to consider all other
request other than the html page?
Thanks in anticipation
Vaibhav
--
View this message in context: http://www.nabble.com/10k-Hits-for-web-site-tp16920984p16920984.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
Re: 10k Hits for web site
Posted by sebb <se...@gmail.com>.
2008/4/28 wvaibhav <va...@sigmainfo.net>:
It's very difficult to tell who wrote what, so please either adjust
you e-mail client so it prefixes lines with >, or don't reply in-line.
>
> > Hello All,
> >
> > I am simulating 10K hits using jmeter distrubuted environment (1 master
> > three slaves) I have planned the task and tried to execute, but having
> > some
> > trouble in execution... and also wanted to confirm the scenario which i
> > have
> > executed....and will be improving gradually
> >
>
> 10K hits in how long?
>
> 10 k in 2000 seconds
>
>
> > I have recorded a script using jmeter proxy. I wanted a rampup period for
> > the execution to be 10 threads per two seconds. So, I selected 50
> > threads,
> > ramp up period 10 and used a loop count for 2000. it takes around 6-7
> > hours
> > to complete the job.... If I increase the number of threads and rampup
> > period and reduce the loop count, i receive out of memory error
>
> where do you get OOM?
>
> JMeter client? JMeter server(s)? all 4? or web server?
>
> Jmeter server only
>
Do you mean the JMeter master? i.e. client?
>
> > and server
> > in distributed environment gets preety slow completion of @ 60% of job
> > say
> > 100 requests in one hour. PF usage is 1.2 GB for 1 GB RAM and cpu usage
> > is
> > around 55 %. How is the scenario?
>
> Which host are you referring to?
>
> PF usage for jmeter server, I am using Jetty server for my web application
>
Do you mean the JMeter master, i.e. client?
>
> > How much rampup period jmeter consider after completion of first loop?
> > whether it considers same rampup period for all the loops?
>
> Ramp-up only applies to first sample in a thread.
>
> The Ramp-up is divided by the number of threads, and each thread is
> delayed by the result.
>
> If rampup is divided by number of threads, then my whole calculation is
> wrong. i have 50 threads, 10 rampup and a loop of 200 for 10000 users. Under
> the assumption that threads are divided by rampup period. I wanted 10000
> threads in 2000 seconds.
If there is one sample in the test plan, then this will generate 10000 samples.
10000 samples in 2000 seconds is 5 samples per second.
You'll need to add a timer; easiest is probably to add a
http://jakarta.apache.org/jmeter/usermanual/component_reference.html#Constant_Throughput_Timer
>
> > What are standard maximum number of threads, rampup period and loop count
> > for heavy loads....
>
> Depends on the test plan, but it should be possible to run several
> hundred threads in one instance of JMeter.
>
> Also depends on how you are running the plan - standalone non-GUI mode
> is cheapest.
>
> I will try it in non gui mode, i donot know how to do it but i will learn
> and execute...
>
It's all described in the documentation:
http://jakarta.apache.org/jmeter/usermanual/get-started.html#non-gui
It's a lot easier than client-server mode...
>
> > Another question is while recording scripts through jmeter, it collects
> > all
> > the image files and other requests from the web page to whom we are
> > having
> > hits. For actual testing whether we are suppose to consider all other
> > request other than the html page?
>
> Depends on what you are trying to test.
>
> Trying to test accessibility of web page for 10000 hits in specified time
> period..
>
Is this to simulate 10000 different users?
In which case you do want to load the images - and other embedded
resources - each time.
> --
> View this message in context: http://www.nabble.com/10k-Hits-for-web-site-tp16920984p16930375.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
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
Re: 10k Hits for web site
Posted by wvaibhav <va...@sigmainfo.net>.
> Hello All,
>
> I am simulating 10K hits using jmeter distrubuted environment (1 master
> three slaves) I have planned the task and tried to execute, but having
> some
> trouble in execution... and also wanted to confirm the scenario which i
> have
> executed....and will be improving gradually
>
10K hits in how long?
10 k in 2000 seconds
> I have recorded a script using jmeter proxy. I wanted a rampup period for
> the execution to be 10 threads per two seconds. So, I selected 50
> threads,
> ramp up period 10 and used a loop count for 2000. it takes around 6-7
> hours
> to complete the job.... If I increase the number of threads and rampup
> period and reduce the loop count, i receive out of memory error
where do you get OOM?
JMeter client? JMeter server(s)? all 4? or web server?
Jmeter server only
> and server
> in distributed environment gets preety slow completion of @ 60% of job
> say
> 100 requests in one hour. PF usage is 1.2 GB for 1 GB RAM and cpu usage
> is
> around 55 %. How is the scenario?
Which host are you referring to?
PF usage for jmeter server, I am using Jetty server for my web application
> How much rampup period jmeter consider after completion of first loop?
> whether it considers same rampup period for all the loops?
Ramp-up only applies to first sample in a thread.
The Ramp-up is divided by the number of threads, and each thread is
delayed by the result.
If rampup is divided by number of threads, then my whole calculation is
wrong. i have 50 threads, 10 rampup and a loop of 200 for 10000 users. Under
the assumption that threads are divided by rampup period. I wanted 10000
threads in 2000 seconds.
> What are standard maximum number of threads, rampup period and loop count
> for heavy loads....
Depends on the test plan, but it should be possible to run several
hundred threads in one instance of JMeter.
Also depends on how you are running the plan - standalone non-GUI mode
is cheapest.
I will try it in non gui mode, i donot know how to do it but i will learn
and execute...
> Another question is while recording scripts through jmeter, it collects
> all
> the image files and other requests from the web page to whom we are
> having
> hits. For actual testing whether we are suppose to consider all other
> request other than the html page?
Depends on what you are trying to test.
Trying to test accessibility of web page for 10000 hits in specified time
period..
--
View this message in context: http://www.nabble.com/10k-Hits-for-web-site-tp16920984p16930375.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
Re: 10k Hits for web site
Posted by sebb <se...@gmail.com>.
2008/4/27 wvaibhav <va...@sigmainfo.net>:
>
> Hello All,
>
> I am simulating 10K hits using jmeter distrubuted environment (1 master
> three slaves) I have planned the task and tried to execute, but having some
> trouble in execution... and also wanted to confirm the scenario which i have
> executed....and will be improving gradually
>
10K hits in how long?
> I have recorded a script using jmeter proxy. I wanted a rampup period for
> the execution to be 10 threads per two seconds. So, I selected 50 threads,
> ramp up period 10 and used a loop count for 2000. it takes around 6-7 hours
> to complete the job.... If I increase the number of threads and rampup
> period and reduce the loop count, i receive out of memory error
where do you get OOM?
JMeter client? JMeter server(s)? all 4? or web server?
> and server
> in distributed environment gets preety slow completion of @ 60% of job say
> 100 requests in one hour. PF usage is 1.2 GB for 1 GB RAM and cpu usage is
> around 55 %. How is the scenario?
Which host are you referring to?
> How much rampup period jmeter consider after completion of first loop?
> whether it considers same rampup period for all the loops?
Ramp-up only applies to first sample in a thread.
The Ramp-up is divided by the number of threads, and each thread is
delayed by the result.
> What are standard maximum number of threads, rampup period and loop count
> for heavy loads....
Depends on the test plan, but it should be possible to run several
hundred threads in one instance of JMeter.
Also depends on how you are running the plan - standalone non-GUI mode
is cheapest.
> Another question is while recording scripts through jmeter, it collects all
> the image files and other requests from the web page to whom we are having
> hits. For actual testing whether we are suppose to consider all other
> request other than the html page?
Depends on what you are trying to test.
> Thanks in anticipation
>
> Vaibhav
> --
> View this message in context: http://www.nabble.com/10k-Hits-for-web-site-tp16920984p16920984.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
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
Re: 10k Hits for web site
Posted by David Brown <da...@davidwbrown.name>.
Hello, i don't know about all of this but what i have done in the past to exclude images: jpeg, gif, etc. is when your created your HTTP Proxy Server in the Workbench to add to the: URL Patterns to Exclude. See the doco at: http://jakarta.apache.org/jmeter/usermanual/jmeter_proxy_step_by_step.pdf . HTH.
wvaibhav wrote ..
>
> Hello All,
>
> I am simulating 10K hits using jmeter distrubuted environment (1 master
> three slaves) I have planned the task and tried to execute, but having some
> trouble in execution... and also wanted to confirm the scenario which i have
> executed....and will be improving gradually
>
> I have recorded a script using jmeter proxy. I wanted a rampup period for
> the execution to be 10 threads per two seconds. So, I selected 50 threads,
> ramp up period 10 and used a loop count for 2000. it takes around 6-7 hours
> to complete the job.... If I increase the number of threads and rampup
> period and reduce the loop count, i receive out of memory error and server
> in distributed environment gets preety slow completion of @ 60% of job say
> 100 requests in one hour. PF usage is 1.2 GB for 1 GB RAM and cpu usage is
> around 55 %. How is the scenario?
> How much rampup period jmeter consider after completion of first loop?
> whether it considers same rampup period for all the loops?
>
> What are standard maximum number of threads, rampup period and loop count
> for heavy loads....
>
> Another question is while recording scripts through jmeter, it collects all
> the image files and other requests from the web page to whom we are having
> hits. For actual testing whether we are suppose to consider all other
> request other than the html page?
>
> Thanks in anticipation
>
> Vaibhav
> --
> View this message in context: http://www.nabble.com/10k-Hits-for-web-site-tp16920984p16920984.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
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org