For Grinder GURU's

[smiles]

Hi,

 

We are using grinder for our performance test and really it is very much realistic tool

 

My issue is, when I run 30 threads with 1000 runs (30000 tests) I ignore around 25 samples.

 

1 sample = ~ 250 tests.--- > this is got from the console by looking in to ‘successful tests/sample’

 

Next while am running for 60 threads with 1000 runs (60000 tests) the test collected per sample is high

i.e.) 1 sample = ~ 450 tests, Also the TPS is around 80 and Service time is around 600

 

I want to know why there is being a difference in collecting samples when the numbers of tests are increased. Also I referred to documents but unable to find the things.

So when the number of test is less the grinder is sending less number of test to the application server and when the tests number is high the grinder is increasing the test send to the server(his shows high TPS) so we cannot measure the full capability of the application server that how much transaction(max) it can process,l This is due to the inconsistency in sample collection mechanism of grinder(hope)

 

Please can anyone give me ideas regarding this issue. Also any document where I can find this

 

Thanks in advance

 

Regards,

Smiles

 

[Philip Aston]

smiles wrote:

> We are using grinder for our performance test and really it is very much
> realistic tool
>
> My issue is, when I run 30 threads with 1000 runs (30000 tests) I ignore
> around 25 samples.
>  
> 1 sample = ~ 250 tests.--- > this is got from the console by looking in to
> ‘successful tests/sample’
>  
>
> Next while am running for 60 threads with 1000 runs (60000 tests) the test
> collected per sample is high
>
> i.e.) 1 sample = ~ 450 tests, Also the TPS is around 80 and Service time is
> around 600
>
> I want to know why there is being a difference in collecting samples when
> the numbers of tests are increased. Also I referred to documents but unable
> to find the things.
>
> So when the number of test is less the grinder is sending less number of
> test to the application server and when the tests number is high the grinder
> is increasing the test send to the server(his shows high TPS) so we cannot
> measure the full capability of the application server that how much
> transaction(max) it can process,l This is due to the inconsistency in sample
> collection mechanism of grinder(hope)
>
>  

Each Sample is based on a fixed time window. This window is used by the
console to calculate derived statistics such as mean and peak TPS.

I see nothing that suggests an inconsistency in the sample model. You
are using more threads in the second scenario, so it is not surprising
that there are more tests completing within a sample window. In the
first scenario, through put is most likely artificially constrained
because each of the 30 threads is waiting for a response and is unable
to send another request until  that response is received. Generally, you
will be able to increase the TPS by increasing the number of threads
until you hit a bottle neck - this bottleneck might be the server, the
capacity of the load injection machines, the network, or something else.

- Phil


Notice:  This email message, together with any attachments, may contain information  of  BEA Systems,  Inc.,  its subsidiaries  and  affiliated entities,  that may be confidential,  proprietary,  copyrighted  and/or legally privileged, and is intended solely for the use of the individual or entity named in this message. If you are not the intended recipient, and have received this message in error, please immediately return this by email and then delete it.

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[smiles]

Hi Philip,

Very much thanks

Ya I agreed with you.
>threads is waiting for a response and is unable to send another request until  that response is received.

So after I get the response from the server for what all the threads I sent to server then only the next set of threads will be processed or will be sent to server. Is it right? So for that much time the console will be on collecting the samples?

Also some time the sample is collecting  ~250 threads on the other time it is collecting ~100 threads next some 70 and then 500 so on......
Why it is making a difference in collecting the sample every time. What factor is affecting this.


If i get the sample collection as constant so i will get steady TPS.

Thanks

Regards,
Smiles


Philip Aston wrote:
>
> smiles wrote:
>> We are using grinder for our performance test and really it is very much
>> realistic tool
>>
>> My issue is, when I run 30 threads with 1000 runs (30000 tests) I ignore
>> around 25 samples.
>>  
>> 1 sample = ~ 250 tests.--- > this is got from the console by looking in
>> to
>> ‘successful tests/sample’
>>  
>>
>> Next while am running for 60 threads with 1000 runs (60000 tests) the
>> test
>> collected per sample is high
>>
>> i.e.) 1 sample = ~ 450 tests, Also the TPS is around 80 and Service time
>> is
>> around 600
>>
>> I want to know why there is being a difference in collecting samples when
>> the numbers of tests are increased. Also I referred to documents but
>> unable
>> to find the things.
>>
>> So when the number of test is less the grinder is sending less number of
>> test to the application server and when the tests number is high the
>> grinder
>> is increasing the test send to the server(his shows high TPS) so we
>> cannot
>> measure the full capability of the application server that how much
>> transaction(max) it can process,l This is due to the inconsistency in
>> sample
>> collection mechanism of grinder(hope)
>>
>>  
>
> Each Sample is based on a fixed time window. This window is used by the
> console to calculate derived statistics such as mean and peak TPS.
>
> I see nothing that suggests an inconsistency in the sample model. You
> are using more threads in the second scenario, so it is not surprising
> that there are more tests completing within a sample window. In the
> first scenario, through put is most likely artificially constrained
> because each of the 30 threads is waiting for a response and is unable
> to send another request until  that response is received. Generally, you
> will be able to increase the TPS by increasing the number of threads
> until you hit a bottle neck - this bottleneck might be the server, the
> capacity of the load injection machines, the network, or something else.
>
> - Phil
>
>
> Notice:  This email message, together with any attachments, may contain
> information  of  BEA Systems,  Inc.,  its subsidiaries  and  affiliated
> entities,  that may be confidential,  proprietary,  copyrighted  and/or
> legally privileged, and is intended solely for the use of the individual
> or entity named in this message. If you are not the intended recipient,
> and have received this message in error, please immediately return this by
> email and then delete it.
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> grinder-use mailing list
> [e-mail]
> https://lists.sourceforge.net/lists/listinfo/grinder-use
>
>
Quoted from:
http://www.nabble.com/For-Grinder-GURU%27s-tp17662100p17665310.html

[Philip Aston-3]

smiles wrote:

> Hi Philip,
>
> Very much thanks
>
> Ya I agreed with you.
>  
>> threads is waiting for a response and is unable to send another request
>> until that response is received.
>>    
> So after I get the response from the server for what all the threads I sent
> to server then only the next set of threads will be processed or will be
> sent to server. Is it right?

Not quite, there's no synchronisation between the threads. The point is
that each thread will process one request at a time, so (until the
server is saturated), you can increase throughput by adding more threads.

> So for that much time the console will be on collecting the samples?
>
> Also some time the sample is collecting  ~250 threads on the other time it
> is collecting ~100 threads next some 70 and then 500 so on......
> Why it is making a difference in collecting the sample every time. What
> factor is affecting this.
>  

This is just chance. When you are seeing more test resultsh (not
"threads") in a sample, its most likely because the threads were mainly
performing cheap requests (e.g. image retrieval) and so could do more.

The sample period is also the frequency at which the console updates its
display, so the default value is deliberately set low to provide a more
immediate display of activity. For formal test runs, I'd advise
increasing the sample period to 10 seconds so that the averages and
peaks are more accurate.

- Phil



-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[Balbir.Kaur]

Hi All,

I am very very new to Jython, Python and Grinder
I am trying to do XML parsing in Grinder script , here are different options I tried -

1. Tried to run amazon.py script from Grinder example and got import error for Javadom
   after doing some google research tried to install PyXml but that only supports Python2.4....I have Python2.5 on my windows machine

2. Same code using SAX , expat which works in Python , gives weird msg in Jython .

Can somebody please share an example of XML parsing and  extracting text   in a grinder script ?

Thanks
-Anu











 
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[VAN GIEL Marc-2]

Hi,

Here is some code I use to parse XML (using Xpath expressions) in Jython (using Java 5):

from java.io import ByteArrayInputStream
from org.xml.sax import InputSource

from javax.xml.parsers import DocumentBuilderFactory
from javax.xml.xpath import XPath, XPathFactory, XPathExpression, XPathConstants
from org.w3c.dom import Document
from org.w3c.dom import Node
from org.w3c.dom import NodeList as XPathNodeList


def findXpathNodes(xmlAsText, xpathExpr):
        inputSrc = InputSource(ByteArrayInputStream(String(xmlAsText).getBytes()))
        dfactory = DocumentBuilderFactory.newInstance()
        dfactory.setNamespaceAware(0)
        dfactory.setValidating(0)
        docBuilder = dfactory.newDocumentBuilder()
        doc = docBuilder.parse(inputSrc)
       
        factory = XPathFactory.newInstance()
        xpath = factory.newXPath()
        expr = xpath.compile(xpathExpr);

        nodeList = expr.evaluate(doc, XPathConstants.NODESET)
        nodeCount = nodeList.getLength()
        i = 0
        result = []
        while i < nodeCount:
            result.append(nodeList.item(i))
            i = i + 1
        return result



Hope this will help you further.

cheers,
Marc.

On Tue, Jun 10, 2008 at 11:22 PM, <Balbir.Kaur@...> wrote:

Hi All,

I am very very new to Jython, Python and Grinder
I am trying to do XML parsing in Grinder script , here are different options I tried -

1. Tried to run amazon.py script from Grinder example and got import error for Javadom
   after doing some google research tried to install PyXml but that only supports Python2.4....I have Python2.5 on my windows machine

2. Same code using SAX , expat which works in Python , gives weird msg in Jython .

Can somebody please share an example of XML parsing and  extracting text   in a grinder script ?

Thanks
-Anu











 
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[Balbir.Kaur]

Thanks for replying :)

Luckily I could run one SAX parser example from 'Python cookbook' in jyhton
my next question is to use third party utility 'Visual Diff' to comapre 2 XML files & trying to automate this process in Grinder...this is not for load testing purpose but just to make the whole process automated with XML output from grinder script ......please do suggest if I need to do different tools for this purpose.

Thanks
-Anu

"Marc Van Giel" <marc.vangiel@...>
Sent by: grinder-use-bounces@...

06/11/2008 12:53 AM

Please respond to
grinder-use <grinder-use@...>

To	grinder-use <grinder-use@...>
cc
Subject	Re: [Grinder-use] XML parsing Jython

Hi,

Here is some code I use to parse XML (using Xpath expressions) in Jython (using Java 5):

from java.io import ByteArrayInputStream
from org.xml.sax import InputSource

from javax.xml.parsers import DocumentBuilderFactory
from javax.xml.xpath import XPath, XPathFactory, XPathExpression, XPathConstants
from org.w3c.dom import Document
from org.w3c.dom import Node
from org.w3c.dom import NodeList as XPathNodeList


def findXpathNodes(xmlAsText, xpathExpr):
       inputSrc = InputSource(ByteArrayInputStream(String(xmlAsText).getBytes()))
       dfactory = DocumentBuilderFactory.newInstance()
       dfactory.setNamespaceAware(0)
       dfactory.setValidating(0)
       docBuilder = dfactory.newDocumentBuilder()
       doc = docBuilder.parse(inputSrc)
       
       factory = XPathFactory.newInstance()
       xpath = factory.newXPath()
       expr = xpath.compile(xpathExpr);

       nodeList = expr.evaluate(doc, XPathConstants.NODESET)
       nodeCount = nodeList.getLength()
       i = 0
       result = []
       while i < nodeCount:
           result.append(nodeList.item(i))
           i = i + 1
       return result



Hope this will help you further.

cheers,
Marc.

On Tue, Jun 10, 2008 at 11:22 PM, <Balbir.Kaur@...> wrote:


Hi All,

I am very very new to Jython, Python and Grinder
I am trying to do XML parsing in Grinder script , here are different options I tried -

1. Tried to run amazon.py script from Grinder example and got import error for Javadom
  after doing some google research tried to install PyXml but that only supports Python2.4....I have Python2.5 on my windows machine

2. Same code using SAX , expat which works in Python , gives weird msg in Jython .

Can somebody please share an example of XML parsing and  extracting text   in a grinder script ?

Thanks
-Anu












--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[Balbir.Kaur]

Hi,
I want to use  'Visual Diff' to comapre  XML files returned from Grinder script, is there a way to automate this process ?
Thanks
-Anu
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[Balbir.Kaur]

Hi Grinder Guru's

My Jython request's output is in form of an image/png ......can I store this to to a file ?

Thanks
-Anu
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[VAN GIEL Marc-2]

Hello,

Sure you can:

response = request.GET("....")
imgPng = response.data

f = open('myImage.pgn','w')
f.write(imgPng)
f.close()



cheers,
Marc.

On Thu, Jun 12, 2008 at 10:41 PM, <Balbir.Kaur@...> wrote:

Hi Grinder Guru's

My Jython request's output is in form of an image/png ......can I store this to to a file ?

Thanks
-Anu
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[Balbir.Kaur]

Marc,

Still it doesn't work .....

here is my code -
 
result=request1.GET("%s?type=image&session=%s&width=%d&height=%d" %(requestURL,session,width,height))
WriteToFile(result.data)
def WriteToFile(imgPng):
   filename=grinder.getFilenameFactory().createFilename("page","-%d.png" %grinder.runNumber)
   file =open(filename,'w')
   #print >>file,imgPng
   file.write(imgPng)
   file.close()
 
 
error msg -
/13/08 8:36:02 AM (thread 0 run 0 test 1): Aborted run due to Jython exception: TypeError: write requires a string as its argument [calling TestRunner]
TypeError: write requires a string as its argument
File "C:\grinder-3.0.1\AIFP1.py", line 78, in WriteToFile
File "C:\grinder-3.0.1\AIFP1.py", line 71, in run
File "C:\grinder-3.0.1\AIFP1.py", line 39, in __call__

 
google results -
Issue 1854873: Patch for [ 1841644 ] Mmap module is not in jython ...

In jython, due to lack of buffer support, functions like file.write() and re.search() accept only a string instance as argument. ...

no idea how to fix these things ....just hoping for a new Jython release with all these patches
-f.write
-javadom - parser problem with Xerces
-filecmp utility

Can somebody help me with implementing the proposed patches in my jython2.2.1 install ?

Thanks
-Anu

"Marc Van Giel" <marc.vangiel@...>
Sent by: grinder-use-bounces@...

06/13/2008 12:17 AM

Please respond to
grinder-use <grinder-use@...>

To	grinder-use <grinder-use@...>
cc
Subject	Re: [Grinder-use] Getting image as output from Jython script

Hello,

Sure you can:

response = request.GET("....")
imgPng = response.data

f = open('myImage.pgn','w')
f.write(imgPng)
f.close()



cheers,
Marc.


On Thu, Jun 12, 2008 at 10:41 PM, <Balbir.Kaur@...> wrote:

Hi Grinder Guru's

My Jython request's output is in form of an image/png ......can I store this to to a file ?

Thanks
-Anu
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[RUSSELL Scott-2]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Balbir,

           You might like to try….

 

File =open(filename, “wb”)

 

       For your information, the “b” writes the file in binary mode(as opposed to text). Also, I am not sure, but I always use quotes(“) not a quote(‘).  Not sure if it actually makes a difference, but you can try it and let us know……

 

                                        Hope that helps.

 

 

                                                              Sc0tt….

 

 

 

---

From: grinder-use-bounces@... [mailto:grinder-use-bounces@...] On Behalf Of Balbir.Kaur@...
Sent: Friday, June 13, 2008 6:50 PM
To: grinder-use
Subject: Re: [Grinder-use] Getting image as output from Jython script
Importance: High

 

Marc,

Still it doesn't work .....

here is my code -
 
result=request1.GET("%s?type=image&session=%s&width=%d&height=%d" %(requestURL,session,width,height))
WriteToFile(result.data)
def WriteToFile(imgPng):
   filename=grinder.getFilenameFactory().createFilename("page","-%d.png" %grinder.runNumber)
   file =open(filename,'w')
   #print >>file,imgPng
   file.write(imgPng)
   file.close()
 
 
error msg -
/13/08 8:36:02 AM (thread 0 run 0 test 1): Aborted run due to Jython exception: TypeError: write requires a string as its argument [calling TestRunner]
TypeError: write requires a string as its argument
File "C:\grinder-3.0.1\AIFP1.py", line 78, in WriteToFile
File "C:\grinder-3.0.1\AIFP1.py", line 71, in run
File "C:\grinder-3.0.1\AIFP1.py", line 39, in __call__

 
google results -
Issue 1854873: Patch for [ 1841644 ] Mmap module is not in jython ...

In jython, due to lack of buffer support, functions like file.write() and re.search() accept only a string instance as argument. ...

no idea how to fix these things ....just hoping for a new Jython release with all these patches
-f.write
-javadom - parser problem with Xerces
-filecmp utility

Can somebody help me with implementing the proposed patches in my jython2.2.1 install ?

Thanks
-Anu

"Marc Van Giel" <marc.vangiel@...>
Sent by: grinder-use-bounces@...

06/13/2008 12:17 AM

Please respond to
grinder-use <[e-mail]>

To	grinder-use <[e-mail]>
cc
Subject	Re: [Grinder-use] Getting image as output from Jython script

 

Hello,

Sure you can:

response = request.GET("....")
imgPng = response.data

f = open('myImage.pgn','w')
f.write(imgPng)
f.close()



cheers,
Marc.


On Thu, Jun 12, 2008 at 10:41 PM, <Balbir.Kaur@...> wrote:

Hi Grinder Guru's

My Jython request's output is in form of an image/png ......can I store this to to a file ?

Thanks
-Anu
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php_______________________________________________
grinder-use mailing list
[e-mail]
https://lists.sourceforge.net/lists/listinfo/grinder-use

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
grinder-use mailing list
grinder-use@...
https://lists.sourceforge.net/lists/listinfo/grinder-use

[RUSSELL Scott-2]

Some parts of this message have been removed. Learn more about Nabble's security policy.