Uneven Distribution of Usage

[Wang, Harry]

I am testing some WSAPIs (12 of them) and the production usage logs
shows 
that 1 of them has over 95% of the requests over some time period. Is
this model typical? 
How does one effectively test the distribution of the requests?
 
My initial thought was to put it into a class all by itself.  Take it as
a abnormally.  That isn't what it is. 
It is a "ReadOnly" API that retrieves info and returns it to the user.
It is pretty typical behavior. 
So how should distribute the number of requests in my perf test plan? 
 
Harry C. Wang
Senior Test Engineer
AIM CDID hwang98109


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

Am I on the right track?

[h b]

Hello, I'm new to JMeter.  I was wondering if you could share some of your
experiences with me.



I'm testing a web application, I want to know how scalable the app is and
how well it performs.



setup: The server & the jmeter-machine are seperate machines.



I have set up a test, then I ramp up the number of threads.  I started with
10 threads 50 requests and worked my way up to 100 threads 50 requests.
ramp up period of 1 second for all test cases.  Then I look at the result
using the Aggregate Report and Graph Result listeners.



During the ramp-up what I'm seeing is that the throughput increases to 17
requests per second and then tapers off.  I'm guessing that this is a good
measure of the scalability of the application.  Is that correct?



The other numbers that I find interesting are the median response time
(+deviation) and the 90% line:  These numbers increase as the number of
threads increase.  My understanding is that when I'm running the case 50
threads, 50 requests what's happening is that for most of the duration of
the test the server is loaded with 50 requests.  So the median response time
of X seems to suggest, that when the server is loaded with 50 requests the
average response time is X.  Is this correct?


The application that I'm testing needs to respond in less than 2 seconds
when it is processing 50-100 requests.  Am I on the right track with my test
setup and am I understanding the results correctly?


Thanks in advance.

Re: simple data writer date

[sebb]

Works fine for me. Which version on JMeter are you using?

On 28/01/2008, Ray Trejo <[EMAIL PROTECTED]> wrote:
> I apologize for the second post on the same issue.
> So here is my issue.
> I am updating the Simple data writer with the below information thru the
> gui.
> In the Filename field.
>
> C:\log${__time()}.jtl
>
> Its actually parsing all the text.
> So the log file is created as:  log${__time()}.jtl
> For some reason it is not grabbing the date and time.
>
> Any suggestions would be greatly appreciated.
>
>
>
> Thank you,
>
> Ray Trejo
> Rackspace Managed Hosting
> Quality Assurance Engineer
> Direct:  (210) 312 - 5326
> Email:  [EMAIL PROTECTED]
> -Original Message-
> From: sebb [mailto:[EMAIL PROTECTED]
> Sent: Monday, January 28, 2008 9:11 AM
> To: JMeter Users List
> Subject: Re: simple data writer date
>
> On 28/01/2008, Ray Trejo <[EMAIL PROTECTED]> wrote:
> > Hi,
> >
> > I would like to insert the current date and time (formatted:  MMDDYY
> > HHMI) into the filename field of the simple data writer.   Is this
> > possible, and how do I do it?
> >
>
> http://jakarta.apache.org/jmeter/usermanual/functions.html#predefinedvar
> s
>
> You can also use functions - e.g. __P() and __time() - in the filename
> field.
>
> >
> >
> > I tried querying the db and formatting the date, using a regex, but
> this
> > did not work.
> >
>
> That's because the log file name is resolved before the test starts.
>
> > The file name displays. Log_${date}.jtl
> >
> >  instead of log_012807 0900.jtl.
> >
> >
> >
> >
> >
> >
> >
> > Thank you,
> >
> >
> >
> > Ray Trejo
> >
> > Rackspace Managed Hosting
> >
> > Quality Assurance Engineer
> >
> > Direct:  (210) 312 - 5326
> >
> > Email:  [EMAIL PROTECTED] 
> >
> >
> >
> >
> >
> > Confidentiality Notice: This e-mail message (including any attached or
> > embedded documents) is intended for the exclusive and confidential use
> of the
> > individual or entity to which this message is addressed, and unless
> otherwise
> > expressly indicated, is confidential and privileged information of
> Rackspace.
> > Any dissemination, distribution or copying of the enclosed material is
> prohibited.
> > If you receive this transmission in error, please notify us
> immediately by e-mail
> > at [EMAIL PROTECTED], and delete the original message.
> > Your cooperation is appreciated.
> >
> >
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
>
> Confidentiality Notice: This e-mail message (including any attached or
> embedded documents) is intended for the exclusive and confidential use of the
> individual or entity to which this message is addressed, and unless otherwise
> expressly indicated, is confidential and privileged information of Rackspace.
> Any dissemination, distribution or copying of the enclosed material is 
> prohibited.
> If you receive this transmission in error, please notify us immediately by 
> e-mail
> at [EMAIL PROTECTED], and delete the original message.
> Your cooperation is appreciated.
>
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

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

RE: simple data writer date

[Ray Trejo]

I apologize for the second post on the same issue.
So here is my issue.
I am updating the Simple data writer with the below information thru the
gui.
In the Filename field.

C:\log${__time()}.jtl

Its actually parsing all the text.
So the log file is created as:  log${__time()}.jtl
For some reason it is not grabbing the date and time.  

Any suggestions would be greatly appreciated.



Thank you,
 
Ray Trejo
Rackspace Managed Hosting
Quality Assurance Engineer
Direct:  (210) 312 - 5326
Email:  [EMAIL PROTECTED]
-Original Message-
From: sebb [mailto:[EMAIL PROTECTED] 
Sent: Monday, January 28, 2008 9:11 AM
To: JMeter Users List
Subject: Re: simple data writer date

On 28/01/2008, Ray Trejo <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I would like to insert the current date and time (formatted:  MMDDYY
> HHMI) into the filename field of the simple data writer.   Is this
> possible, and how do I do it?
>

http://jakarta.apache.org/jmeter/usermanual/functions.html#predefinedvar
s

You can also use functions - e.g. __P() and __time() - in the filename
field.

>
>
> I tried querying the db and formatting the date, using a regex, but
this
> did not work.
>

That's because the log file name is resolved before the test starts.

> The file name displays. Log_${date}.jtl
>
>  instead of log_012807 0900.jtl.
>
>
>
>
>
>
>
> Thank you,
>
>
>
> Ray Trejo
>
> Rackspace Managed Hosting
>
> Quality Assurance Engineer
>
> Direct:  (210) 312 - 5326
>
> Email:  [EMAIL PROTECTED] 
>
>
>
>
>
> Confidentiality Notice: This e-mail message (including any attached or
> embedded documents) is intended for the exclusive and confidential use
of the
> individual or entity to which this message is addressed, and unless
otherwise
> expressly indicated, is confidential and privileged information of
Rackspace.
> Any dissemination, distribution or copying of the enclosed material is
prohibited.
> If you receive this transmission in error, please notify us
immediately by e-mail
> at [EMAIL PROTECTED], and delete the original message.
> Your cooperation is appreciated.
>
>

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



Confidentiality Notice: This e-mail message (including any attached or
embedded documents) is intended for the exclusive and confidential use of the
individual or entity to which this message is addressed, and unless otherwise
expressly indicated, is confidential and privileged information of Rackspace. 
Any dissemination, distribution or copying of the enclosed material is 
prohibited.
If you receive this transmission in error, please notify us immediately by 
e-mail
at [EMAIL PROTECTED], and delete the original message. 
Your cooperation is appreciated.


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

Re: Time spent where?

[Lists]

They show on average response times practically no difference between apache
and jmeter. i.e. 1-2ms or thereabouts.

- Original Message 
From: JMeter Users List 
To: JMeter Users List , Lists
<[EMAIL PROTECTED]>
Subject: Re: Time spent where?
Date: 28/01/08 17:03

> On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> > This difference is a huge one for our system.  We achieve average
response
> > times under normal load of 30-40ms end to end.
> 
> What do the apache logs show in this situation?
> 
> > I dont think this is network contention.  nicstat shows us the
interface
> > rarely gets above 20% utilisation.
> >
> > However i shall get the network admin guy to setup some logging so we
can
> > prove this.
> >
> > I will also look at apache stats and see how many mb's are
transferred
> > during the 10min test.
> > - Original Message 
> > From: sebb <[EMAIL PROTECTED]>
> > To: JMeter Users List , Lists
> > <[EMAIL PROTECTED]>
> > Subject: Re: Time spent where?
> > Date: 28/01/08 16:48
> >
> > > On 28/01/2008, Lists <[EMAIL PROTECTED]>
wrote:
> > > > >
> > > > > How much longer? A few percent or much more?
> > > > >
> > > >
> > > > significantly.  an average of 89ms at the apache logs,
and yet 140 at
> > >
> > > Does not seem too bad considering that the data has to be
transferred
> > > across the network.
> > >
> > > > jmeter.  The key point is though that this latency is
pratically zero
> > at
> > > > lower loads.  The reason it causes us a problem is we
define failure
> > by
> > > > saying 95% of responses must be less than a given
response time - so
> > this
> > > > extra gap is seemingly causing us to fail the tests,
but
> > unnecessarily.
> > >
> > > Depends on the target of course.
> > >
> > > > > Are all samples affected?
> > > >
> > > > Err, good question, i've been looking at averages, so i
dont know at
> > this
> > > > point.  I guess i should look at standard deviation
perhaps..?
> > > >
> > > > > Does the additional time depend on the sample
size?
> > > >
> > > > Err, sample size?  or throughput?  yes it increases
with throughput.
> > > >
> > >
> > > I meant the number of bytes downloaded.
> > >
> > > Bigger samples generally take longer to download.
> > > Some conditions (e.g. network contention) may affect larger
samples
> > > more than smaller ones as the transfer time is greater as a
proportion
> > > of total time.
> > >
> > > > > Note that the elapsed time will necessarily be
longer than the
> > apache
> > > > > log time; the JMeter latency should be closer
to the apache log
> > > > > timings.
> > > > >
> > > > > Are you using any functions in the samplers?
Function execution
> > is
> > > > > single-threaded.
> > > >
> > > > I'm not sure, i'll double check.  However i have
profiled jmeter
> > recently
> > > > and it didnt seem to be blocking - i do need to double
check this,
> > hopefully
> > > > i'll do that tomorrow.
> > > >
> > > >
> > > > > &gt; Thanks,
> > > > > &gt; Dan
> > > > > &gt;
> > > > > &gt;

> > > > > &gt; Message sent using UebiMiau
2.7.10
> > > > > &gt;
> > > > > &gt;
> > > > > &gt;
> > > > > &gt;
> > -
> > > > > &gt; To unsubscribe, e-mail:
> > [EMAIL PROTECTED]
> > > > > &gt; For additional commands, e-mail:
> > [EMAIL PROTECTED]
> > > > > &gt;
> > > > > &gt;
> > > > >
> > > > >
> > -
> > > > > To unsubscribe, e-mail:
> > [EMAIL PROTECTED]
> > > > > For additional commands, e-mail:
> > [EMAIL PROTECTED]
> > > > >
> > > > >
> > > > >
> > > >
> > > > 
> > > > Message sent using UebiMiau 2.7.10
> > > >
> > > >
> > > >
> > > >
-
> > > > To unsubscribe, e-mail:
[EMAIL PROTECTED]
> > > > For additional commands, e-mail:
[EMAIL PROTECTED]
> > > >
> > > >
> > >
> > >
> >
> > 
> > Message sent using UebiMiau 2.7.10
> >
> >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
> 
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 
> 


Message sent using UebiMiau 2.7.10



-
To unsubscr

Re: Time spent where?

[sebb]

On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> This difference is a huge one for our system.  We achieve average response
> times under normal load of 30-40ms end to end.

What do the apache logs show in this situation?

> I dont think this is network contention.  nicstat shows us the interface
> rarely gets above 20% utilisation.
>
> However i shall get the network admin guy to setup some logging so we can
> prove this.
>
> I will also look at apache stats and see how many mb's are transferred
> during the 10min test.
> - Original Message 
> From: sebb <[EMAIL PROTECTED]>
> To: JMeter Users List , Lists
> <[EMAIL PROTECTED]>
> Subject: Re: Time spent where?
> Date: 28/01/08 16:48
>
> > On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> > > >
> > > > How much longer? A few percent or much more?
> > > >
> > >
> > > significantly.  an average of 89ms at the apache logs, and yet 140 at
> >
> > Does not seem too bad considering that the data has to be transferred
> > across the network.
> >
> > > jmeter.  The key point is though that this latency is pratically zero
> at
> > > lower loads.  The reason it causes us a problem is we define failure
> by
> > > saying 95% of responses must be less than a given response time - so
> this
> > > extra gap is seemingly causing us to fail the tests, but
> unnecessarily.
> >
> > Depends on the target of course.
> >
> > > > Are all samples affected?
> > >
> > > Err, good question, i've been looking at averages, so i dont know at
> this
> > > point.  I guess i should look at standard deviation perhaps..?
> > >
> > > > Does the additional time depend on the sample size?
> > >
> > > Err, sample size?  or throughput?  yes it increases with throughput.
> > >
> >
> > I meant the number of bytes downloaded.
> >
> > Bigger samples generally take longer to download.
> > Some conditions (e.g. network contention) may affect larger samples
> > more than smaller ones as the transfer time is greater as a proportion
> > of total time.
> >
> > > > Note that the elapsed time will necessarily be longer than the
> apache
> > > > log time; the JMeter latency should be closer to the apache log
> > > > timings.
> > > >
> > > > Are you using any functions in the samplers? Function execution
> is
> > > > single-threaded.
> > >
> > > I'm not sure, i'll double check.  However i have profiled jmeter
> recently
> > > and it didnt seem to be blocking - i do need to double check this,
> hopefully
> > > i'll do that tomorrow.
> > >
> > >
> > > > > Thanks,
> > > > > Dan
> > > > >
> > > > > 
> > > > > Message sent using UebiMiau 2.7.10
> > > > >
> > > > >
> > > > >
> > > > >
> -
> > > > > To unsubscribe, e-mail:
> [EMAIL PROTECTED]
> > > > > For additional commands, e-mail:
> [EMAIL PROTECTED]
> > > > >
> > > > >
> > > >
> > > >
> -
> > > > To unsubscribe, e-mail:
> [EMAIL PROTECTED]
> > > > For additional commands, e-mail:
> [EMAIL PROTECTED]
> > > >
> > > >
> > > >
> > >
> > > 
> > > Message sent using UebiMiau 2.7.10
> > >
> > >
> > >
> > > -
> > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > For additional commands, e-mail: [EMAIL PROTECTED]
> > >
> > >
> >
> >
>
> 
> Message sent using UebiMiau 2.7.10
>
>
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

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

Re: Time spent where?

[Lists]

This difference is a huge one for our system.  We achieve average response
times under normal load of 30-40ms end to end.

I dont think this is network contention.  nicstat shows us the interface
rarely gets above 20% utilisation.

However i shall get the network admin guy to setup some logging so we can
prove this.

I will also look at apache stats and see how many mb's are transferred
during the 10min test.

- Original Message 
From: sebb <[EMAIL PROTECTED]>
To: JMeter Users List , Lists
<[EMAIL PROTECTED]>
Subject: Re: Time spent where?
Date: 28/01/08 16:48

> On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> > >
> > > How much longer? A few percent or much more?
> > >
> >
> > significantly.  an average of 89ms at the apache logs, and yet 140 at
> 
> Does not seem too bad considering that the data has to be transferred
> across the network.
> 
> > jmeter.  The key point is though that this latency is pratically zero
at
> > lower loads.  The reason it causes us a problem is we define failure
by
> > saying 95% of responses must be less than a given response time - so
this
> > extra gap is seemingly causing us to fail the tests, but
unnecessarily.
> 
> Depends on the target of course.
> 
> > > Are all samples affected?
> >
> > Err, good question, i've been looking at averages, so i dont know at
this
> > point.  I guess i should look at standard deviation perhaps..?
> >
> > > Does the additional time depend on the sample size?
> >
> > Err, sample size?  or throughput?  yes it increases with throughput.
> >
> 
> I meant the number of bytes downloaded.
> 
> Bigger samples generally take longer to download.
> Some conditions (e.g. network contention) may affect larger samples
> more than smaller ones as the transfer time is greater as a proportion
> of total time.
> 
> > > Note that the elapsed time will necessarily be longer than the
apache
> > > log time; the JMeter latency should be closer to the apache log
> > > timings.
> > >
> > > Are you using any functions in the samplers? Function execution
is
> > > single-threaded.
> >
> > I'm not sure, i'll double check.  However i have profiled jmeter
recently
> > and it didnt seem to be blocking - i do need to double check this,
hopefully
> > i'll do that tomorrow.
> >
> >
> > > > Thanks,
> > > > Dan
> > > >
> > > > 
> > > > Message sent using UebiMiau 2.7.10
> > > >
> > > >
> > > >
> > > >
-
> > > > To unsubscribe, e-mail:
[EMAIL PROTECTED]
> > > > For additional commands, e-mail:
[EMAIL PROTECTED]
> > > >
> > > >
> > >
> > >
-
> > > To unsubscribe, e-mail:
[EMAIL PROTECTED]
> > > For additional commands, e-mail:
[EMAIL PROTECTED]
> > >
> > >
> > >
> >
> > 
> > Message sent using UebiMiau 2.7.10
> >
> >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
> 
> 


Message sent using UebiMiau 2.7.10



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

Re: Definitions for the Deviation

[sebb]

Standard Deviation

On 28/01/2008, Wang, Harry <[EMAIL PROTECTED]> wrote:
> What exactly is the Deviation value in the Graph Results listener?
>
> Is it in the ms (millisecond) value for 1 Standard Deviation or
> something else different?
>
> Harry
>

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

Re: Time spent where?

[sebb]

On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> >
> > How much longer? A few percent or much more?
> >
>
> significantly.  an average of 89ms at the apache logs, and yet 140 at

Does not seem too bad considering that the data has to be transferred
across the network.

> jmeter.  The key point is though that this latency is pratically zero at
> lower loads.  The reason it causes us a problem is we define failure by
> saying 95% of responses must be less than a given response time - so this
> extra gap is seemingly causing us to fail the tests, but unnecessarily.

Depends on the target of course.

> > Are all samples affected?
>
> Err, good question, i've been looking at averages, so i dont know at this
> point.  I guess i should look at standard deviation perhaps..?
>
> > Does the additional time depend on the sample size?
>
> Err, sample size?  or throughput?  yes it increases with throughput.
>

I meant the number of bytes downloaded.

Bigger samples generally take longer to download.
Some conditions (e.g. network contention) may affect larger samples
more than smaller ones as the transfer time is greater as a proportion
of total time.

> > Note that the elapsed time will necessarily be longer than the apache
> > log time; the JMeter latency should be closer to the apache log
> > timings.
> >
> > Are you using any functions in the samplers? Function execution is
> > single-threaded.
>
> I'm not sure, i'll double check.  However i have profiled jmeter recently
> and it didnt seem to be blocking - i do need to double check this, hopefully
> i'll do that tomorrow.
>
>
> > > Thanks,
> > > Dan
> > >
> > > 
> > > Message sent using UebiMiau 2.7.10
> > >
> > >
> > >
> > > -
> > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > For additional commands, e-mail: [EMAIL PROTECTED]
> > >
> > >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
> >
>
> 
> Message sent using UebiMiau 2.7.10
>
>
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

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

Definitions for the Deviation

[Wang, Harry]

What exactly is the Deviation value in the Graph Results listener?  
 
Is it in the ms (millisecond) value for 1 Standard Deviation or
something else different?
 
Harry 

Re: Time spent where?

[Lists]

We already have that data available.  The increase in end to end response
times is not matched identically by an increase in time spent running
queries.  ( There is some increase in db time, but not as much as is seen
end to end )

- Original Message 
From: JMeter Users List 
To: JMeter Users List 
Subject: Re: Time spent where?
Date: 28/01/08 14:58

> what I would do is added some simple code to time how long the queries
take
> and write that to a file.
> 
> Back when I worked for Superpages.com, we had all sorts of performance
> logging, so that we could generate reports every night. In my case, we had
> service level agreements, so we had to log extensively and make sure
> response time were within the requirements.
> 
> peter
> 
> On Jan 28, 2008 9:23 AM, sebb  wrote:
> 
> > On 28/01/2008, Lists  wrote:
> > > Hi,
> > >
> > > I've posted about this before, but got some more info now.
> > >
> > > We have a standard apache -> tomcat -> oracle system.  In the db we
can
> > > access the total time spent in tomcat and oracle.  By analysing the
> > apache
> > > logs (modjk gives you an elapsed time) we can log the time spent in
> > apache,
> > > tomcat and oracle.
> > >
> > > And then of course jmeter gives us the overall end to end response
> > times.
> > >
> > > The problem i have is that at higher loads, the gap of response times
> > > reported by jmeter grows much faster than it does when reported at the
> > > apache layer.  So i thought this was caused by a busy jmeter server.
> > >
> > > So i ran 2 jmeters on 2 separate servers independantly (At half load).
> > > Unfortunately the gap remains.
> > >
> > > Is this a usual scenario that people see at high loads?  could the
> > figures
> > > being reported be lieing to me?  The DB server is under load at this
> > point
> > > but none of the other servers seem stressed.  I'm a bit stuck in how
to
> > > debug this further?
> > >
> >
> > How much longer? A few percent or much more?
> >
> > Are all samples affected?
> >
> > Does the additional time depend on the sample size?
> >
> > Note that the elapsed time will necessarily be longer than the apache
> > log time; the JMeter latency should be closer to the apache log
> > timings.
> >
> > Are you using any functions in the samplers? Function execution is
> > single-threaded.
> >
> > > Thanks,
> > > Dan
> > >
> > > 
> > > Message sent using UebiMiau 2.7.10
> > >
> > >
> > >
> > > -
> > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > For additional commands, e-mail: [EMAIL PROTECTED]
> > >
> > >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
> 


Message sent using UebiMiau 2.7.10



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

Re: Time spent where?

[Lists]

Some further stats; We're run up to about 400 requests per second, on a dual
cpu dual core server.  We're using 405 threads.  1gb of memory is allocated
to jmeter, it doesnt appear to want any more, but plenty more is available
if required.



- Original Message 
From: JMeter Users List 
To: JMeter Users List 
Subject: Re: Time spent where?
Date: 28/01/08 16:17

> > 
> > How much longer? A few percent or much more?
> > 
> 
> significantly.  an average of 89ms at the apache logs, and yet 140 at
> jmeter.  The key point is though that this latency is pratically zero at
> lower loads.  The reason it causes us a problem is we define failure by
> saying 95% of responses must be less than a given response time - so this
> extra gap is seemingly causing us to fail the tests, but unnecessarily.
> 
> > Are all samples affected?
> 
> Err, good question, i've been looking at averages, so i dont know at this
> point.  I guess i should look at standard deviation perhaps..?
> 
> > Does the additional time depend on the sample size?
> 
> Err, sample size?  or throughput?  yes it increases with throughput.
> 
> > Note that the elapsed time will necessarily be longer than the apache
> > log time; the JMeter latency should be closer to the apache log
> > timings.
> > 
> > Are you using any functions in the samplers? Function execution is
> > single-threaded.
> 
> I'm not sure, i'll double check.  However i have profiled jmeter recently
> and it didnt seem to be blocking - i do need to double check this,
hopefully
> i'll do that tomorrow.
> 
> 
> > > Thanks,
> > > Dan
> > >
> > > 
> > > Message sent using UebiMiau 2.7.10
> > >
> > >
> > >
> > >
-
> > > To unsubscribe, e-mail:
[EMAIL PROTECTED]
> > > For additional commands, e-mail:
[EMAIL PROTECTED]
> > >
> > >
> > 
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> > 
> > 
> > 
> 
> 
> Message sent using UebiMiau 2.7.10
> 
> 
> 
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 
> 


Message sent using UebiMiau 2.7.10



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

Re: Time spent where?

[Lists]

> 
> How much longer? A few percent or much more?
> 

significantly.  an average of 89ms at the apache logs, and yet 140 at
jmeter.  The key point is though that this latency is pratically zero at
lower loads.  The reason it causes us a problem is we define failure by
saying 95% of responses must be less than a given response time - so this
extra gap is seemingly causing us to fail the tests, but unnecessarily.

> Are all samples affected?

Err, good question, i've been looking at averages, so i dont know at this
point.  I guess i should look at standard deviation perhaps..?

> Does the additional time depend on the sample size?

Err, sample size?  or throughput?  yes it increases with throughput.

> Note that the elapsed time will necessarily be longer than the apache
> log time; the JMeter latency should be closer to the apache log
> timings.
> 
> Are you using any functions in the samplers? Function execution is
> single-threaded.

I'm not sure, i'll double check.  However i have profiled jmeter recently
and it didnt seem to be blocking - i do need to double check this, hopefully
i'll do that tomorrow.


> > Thanks,
> > Dan
> >
> > 
> > Message sent using UebiMiau 2.7.10
> >
> >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
> 
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 
> 


Message sent using UebiMiau 2.7.10



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

Re: simple data writer date

[sebb]

On 28/01/2008, Ray Trejo <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I would like to insert the current date and time (formatted:  MMDDYY
> HHMI) into the filename field of the simple data writer.   Is this
> possible, and how do I do it?
>

http://jakarta.apache.org/jmeter/usermanual/functions.html#predefinedvars

You can also use functions - e.g. __P() and __time() - in the filename field.

>
>
> I tried querying the db and formatting the date, using a regex, but this
> did not work.
>

That's because the log file name is resolved before the test starts.

> The file name displays. Log_${date}.jtl
>
>  instead of log_012807 0900.jtl.
>
>
>
>
>
>
>
> Thank you,
>
>
>
> Ray Trejo
>
> Rackspace Managed Hosting
>
> Quality Assurance Engineer
>
> Direct:  (210) 312 - 5326
>
> Email:  [EMAIL PROTECTED] 
>
>
>
>
>
> Confidentiality Notice: This e-mail message (including any attached or
> embedded documents) is intended for the exclusive and confidential use of the
> individual or entity to which this message is addressed, and unless otherwise
> expressly indicated, is confidential and privileged information of Rackspace.
> Any dissemination, distribution or copying of the enclosed material is 
> prohibited.
> If you receive this transmission in error, please notify us immediately by 
> e-mail
> at [EMAIL PROTECTED], and delete the original message.
> Your cooperation is appreciated.
>
>

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

simple data writer date

[Ray Trejo]

Hi,

I would like to insert the current date and time (formatted:  MMDDYY
HHMI) into the filename field of the simple data writer.   Is this
possible, and how do I do it?

 

I tried querying the db and formatting the date, using a regex, but this
did not work.

The file name displays. Log_${date}.jtl

 instead of log_012807 0900.jtl.

 

 

 

Thank you,

 

Ray Trejo

Rackspace Managed Hosting

Quality Assurance Engineer

Direct:  (210) 312 - 5326

Email:  [EMAIL PROTECTED]  

 



Confidentiality Notice: This e-mail message (including any attached or
embedded documents) is intended for the exclusive and confidential use of the
individual or entity to which this message is addressed, and unless otherwise
expressly indicated, is confidential and privileged information of Rackspace. 
Any dissemination, distribution or copying of the enclosed material is 
prohibited.
If you receive this transmission in error, please notify us immediately by 
e-mail
at [EMAIL PROTECTED], and delete the original message. 
Your cooperation is appreciated.

Re: Time spent where?

[Peter Lin]

what I would do is added some simple code to time how long the queries take
and write that to a file.

Back when I worked for Superpages.com, we had all sorts of performance
logging, so that we could generate reports every night. In my case, we had
service level agreements, so we had to log extensively and make sure
response time were within the requirements.

peter

On Jan 28, 2008 9:23 AM, sebb <[EMAIL PROTECTED]> wrote:

> On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> > Hi,
> >
> > I've posted about this before, but got some more info now.
> >
> > We have a standard apache -> tomcat -> oracle system.  In the db we can
> > access the total time spent in tomcat and oracle.  By analysing the
> apache
> > logs (modjk gives you an elapsed time) we can log the time spent in
> apache,
> > tomcat and oracle.
> >
> > And then of course jmeter gives us the overall end to end response
> times.
> >
> > The problem i have is that at higher loads, the gap of response times
> > reported by jmeter grows much faster than it does when reported at the
> > apache layer.  So i thought this was caused by a busy jmeter server.
> >
> > So i ran 2 jmeters on 2 separate servers independantly (At half load).
> > Unfortunately the gap remains.
> >
> > Is this a usual scenario that people see at high loads?  could the
> figures
> > being reported be lieing to me?  The DB server is under load at this
> point
> > but none of the other servers seem stressed.  I'm a bit stuck in how to
> > debug this further?
> >
>
> How much longer? A few percent or much more?
>
> Are all samples affected?
>
> Does the additional time depend on the sample size?
>
> Note that the elapsed time will necessarily be longer than the apache
> log time; the JMeter latency should be closer to the apache log
> timings.
>
> Are you using any functions in the samplers? Function execution is
> single-threaded.
>
> > Thanks,
> > Dan
> >
> > 
> > Message sent using UebiMiau 2.7.10
> >
> >
> >
> > -
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

Re: Time spent where?

[sebb]

On 28/01/2008, Lists <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I've posted about this before, but got some more info now.
>
> We have a standard apache -> tomcat -> oracle system.  In the db we can
> access the total time spent in tomcat and oracle.  By analysing the apache
> logs (modjk gives you an elapsed time) we can log the time spent in apache,
> tomcat and oracle.
>
> And then of course jmeter gives us the overall end to end response times.
>
> The problem i have is that at higher loads, the gap of response times
> reported by jmeter grows much faster than it does when reported at the
> apache layer.  So i thought this was caused by a busy jmeter server.
>
> So i ran 2 jmeters on 2 separate servers independantly (At half load).
> Unfortunately the gap remains.
>
> Is this a usual scenario that people see at high loads?  could the figures
> being reported be lieing to me?  The DB server is under load at this point
> but none of the other servers seem stressed.  I'm a bit stuck in how to
> debug this further?
>

How much longer? A few percent or much more?

Are all samples affected?

Does the additional time depend on the sample size?

Note that the elapsed time will necessarily be longer than the apache
log time; the JMeter latency should be closer to the apache log
timings.

Are you using any functions in the samplers? Function execution is
single-threaded.

> Thanks,
> Dan
>
> 
> Message sent using UebiMiau 2.7.10
>
>
>
> -
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

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

Time spent where?

[Lists]

Hi,

I've posted about this before, but got some more info now.

We have a standard apache -> tomcat -> oracle system.  In the db we can
access the total time spent in tomcat and oracle.  By analysing the apache
logs (modjk gives you an elapsed time) we can log the time spent in apache,
tomcat and oracle.

And then of course jmeter gives us the overall end to end response times.

The problem i have is that at higher loads, the gap of response times
reported by jmeter grows much faster than it does when reported at the
apache layer.  So i thought this was caused by a busy jmeter server.

So i ran 2 jmeters on 2 separate servers independantly (At half load). 
Unfortunately the gap remains.

Is this a usual scenario that people see at high loads?  could the figures
being reported be lieing to me?  The DB server is under load at this point
but none of the other servers seem stressed.  I'm a bit stuck in how to
debug this further?

Thanks,
Dan


Message sent using UebiMiau 2.7.10



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

Need info

[J, Rajeev]

All,

 

I am not sure if this is a right place to post this message.

I would like to know if anyone of you are using Jmeter and residing in
Bangalore, India. If so can you please let me know.

Thanks for your support and understanding.

 

Thanks,

Rajeev

 

|First Indian Corporation Pvt. Ltd| Associate Project Manager |
*:[EMAIL PROTECTED] | *:080-44332211 Ext: 385|

VOIP: 714-250-7092|Cell: 94485-06088| 3rd Floor, Block A| Lake View,
Bagmane Tech park| Bangalore 560 093| 

 


**
This message contains confidential information intended only for the use of the 
addressee(s) named above and may contain information that is legally 
privileged.  If you are not the addressee, or the person responsible for 
delivering it to the addressee, you are hereby notified that reading, 
disseminating, distributing or copying this message is strictly prohibited.  If 
you have received this message by mistake, please immediately notify us by 
replying to the message and delete the original message immediately thereafter.

Thank you.

   FADLD Tag
**