Topic
  • 16 replies
  • Latest Post - ‏2013-11-20T17:42:44Z by RFTworker
RFTworker
RFTworker
13 Posts

Pinned topic More time on Page Performance report

‏2013-11-11T19:16:47Z |

Hi,

We have started using Rational performance tester recently. We have  RPT 8.2.0.6 version. I have recored a simple http test, added to a schedule and ran it for 50 users. on some of the pages,in page performance report, it was showing as 23 seconds, (around 23,222 ms). usually the pages load pretty quick and not sure why some pages are showing as more time.

Even some of them are 60 seconds :(. I have installed RPT on my desktop and the application is from the different server. so when I record, I access the url and navigate through the pages. Do I have to configure any kind of connections , if I my application is on a different server ? just like in RFT, I configure my browser and application.....

I just access that server with the URL which has the server name, please let me know if I have to mention that in the locations some where if it is other then local host? or else what could be the problem?? please let me know.

Your help is greatly appreciated!! please let me know if you need more details.

 

Thanks

 

  • RFTworker
    RFTworker
    13 Posts
    ACCEPTED ANSWER

    Re: More time on Page Performance report

    ‏2013-11-20T17:42:44Z  
    • RFTworker
    • ‏2013-11-19T15:04:38Z

    Thanks for your notes and suggestions. they are really helping me.

    I just recorded a new test with one user and played back and saw that the first page was showing 14 seconds which is not correct.

    Also I did not see any data correlation errors, or any warnings or any errors in the error log.

    any ideas?

    I have contacted IBM on this issue and it got resolved. We followed some tips from IBM for setting up HTTP tests and schedules.  Some of them are below:

    1. Rate of change is set to one user for every 3 seconds, instead of all users over
    2. Statistics sample interval is set to 5 seconds, instead of 15 seconds to retrieve the data from the server
    3. Whenever the test play starts automatically selected to clear cookie
    4. Cleared page cache when recording and when test starts
    5. Disabled page cache emulation in tests
    6. Forced the garbage collection manually to know the size the memory for the host controller machine.

    Thank you very much for your support and continuous response!! Very much Appreciated

     

    Thanks

     

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-12T22:16:51Z  

    HI, looks like there's two issues here

    1. long page times being reported

    When running with 50 users it's expected pages would take longer to load than if running 1 user.

    You might try creating a schedule with various stages of say 5, 10, 25, 50 users and see how the page response times decrease with higher user loads

    2. the answer to question about what if the application is on another server?  If you mean the web application is hosted on a machine other than the one RPT workbenh is installed, this is almost always the case

    You don't have to do anything special to record in such an environment most of the time. Depending on the application being recorded you may have to import the RPT supplied certificate when using SSL (https:\\) .  You may have to end up using the Proxy vs the Socks recorder.

    I highly recommend upgrading to the latest version of RPT which is 8502 rather than using 8.2.0.6

     

    Updated on 2013-11-12T22:17:39Z at 2013-11-12T22:17:39Z by PaulLiskay
  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-13T13:52:46Z  

    HI, looks like there's two issues here

    1. long page times being reported

    When running with 50 users it's expected pages would take longer to load than if running 1 user.

    You might try creating a schedule with various stages of say 5, 10, 25, 50 users and see how the page response times decrease with higher user loads

    2. the answer to question about what if the application is on another server?  If you mean the web application is hosted on a machine other than the one RPT workbenh is installed, this is almost always the case

    You don't have to do anything special to record in such an environment most of the time. Depending on the application being recorded you may have to import the RPT supplied certificate when using SSL (https:\\) .  You may have to end up using the Proxy vs the Socks recorder.

    I highly recommend upgrading to the latest version of RPT which is 8502 rather than using 8.2.0.6

     

    Thanks for your reply. I will try to set the users 5,10,25 and will see. I have upgraded RPT to 8.3 version today.

     

    Yes I mean to say that RPT is installed on my Local and I am trying to access a URL from a different server. Can you please explain what you mean by importing Certificate ? also what do you mean by ending up with Proxy vs socks recorder.

     

    please your help is appreciated....I used to work on RFT a lot, but started working on RPT also recently.

     

    Thanks

     

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-13T14:01:35Z  
    • RFTworker
    • ‏2013-11-13T13:52:46Z

    Thanks for your reply. I will try to set the users 5,10,25 and will see. I have upgraded RPT to 8.3 version today.

     

    Yes I mean to say that RPT is installed on my Local and I am trying to access a URL from a different server. Can you please explain what you mean by importing Certificate ? also what do you mean by ending up with Proxy vs socks recorder.

     

    please your help is appreciated....I used to work on RFT a lot, but started working on RPT also recently.

     

    Thanks

     

    Here's a good link on using the new major domo load generation agent now that you're on 8.3

    http://www.ibm.com/developerworks/rational/library/load-generation-agent-performance-testing/

    This link will describe best practices for setting up the topology for high volume tests

    http://www.ibm.com/developerworks/rational/library/optimize-load-handling/index.html

    If recording against a SSL server you'll need to import the RPT SSL certificate so the recording will work, this describes it for IE

    http://www-01.ibm.com/support/docview.wss?uid=swg21372997

    At record time there's an option to choose the type of Proxy recorder, sometimes the SOCKS option is needed, especially for web apps that have a lot of java applets, see attached .jpg

    It's always best to use the defaults in RPT unless there's a known reason not to

     

     

     

    Attachments

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-13T20:28:09Z  

    Here's a good link on using the new major domo load generation agent now that you're on 8.3

    http://www.ibm.com/developerworks/rational/library/load-generation-agent-performance-testing/

    This link will describe best practices for setting up the topology for high volume tests

    http://www.ibm.com/developerworks/rational/library/optimize-load-handling/index.html

    If recording against a SSL server you'll need to import the RPT SSL certificate so the recording will work, this describes it for IE

    http://www-01.ibm.com/support/docview.wss?uid=swg21372997

    At record time there's an option to choose the type of Proxy recorder, sometimes the SOCKS option is needed, especially for web apps that have a lot of java applets, see attached .jpg

    It's always best to use the defaults in RPT unless there's a known reason not to

     

     

     

    Thanks a lot for your links.

    The URL that I access to,  also get certification error, asking do you want to continue and I will click continue and will proceed. The steps that are mentioned on the above Tutorial did not work for me to add my site to trusted sites. Also I did not get  any certification error when I recorded or played back my script.

    The Only issue I had is when I ran for 50 users, each page response time was way high (some of them are 60 seconds), and also another thing that happens is all my Log in ID's that I used in the test get locked or stuck and I had to restart my server in order them to reuse.

    Which is the best way to set up 50 users?...in User load, Think time....what are the best options that I can select?

    Can I specify 50 users in User load in one Stage,,,, run a schedule with 2 tests in 2 diff user loads(50% each)?  I am not sure if I am setting them correctly.

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-13T20:32:00Z  
    • RFTworker
    • ‏2013-11-13T20:28:09Z

    Thanks a lot for your links.

    The URL that I access to,  also get certification error, asking do you want to continue and I will click continue and will proceed. The steps that are mentioned on the above Tutorial did not work for me to add my site to trusted sites. Also I did not get  any certification error when I recorded or played back my script.

    The Only issue I had is when I ran for 50 users, each page response time was way high (some of them are 60 seconds), and also another thing that happens is all my Log in ID's that I used in the test get locked or stuck and I had to restart my server in order them to reuse.

    Which is the best way to set up 50 users?...in User load, Think time....what are the best options that I can select?

    Can I specify 50 users in User load in one Stage,,,, run a schedule with 2 tests in 2 diff user loads(50% each)?  I am not sure if I am setting them correctly.

    I am attaching the screen shots for the set up that I have tried with my schedule and also the report that I got. And as I said that everytime after I run the schedule my Test Id's that I used gets stuck :(

     

    Any help is greatly appreciated

    Attachments

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-13T22:32:21Z  
    • RFTworker
    • ‏2013-11-13T20:32:00Z

    I am attaching the screen shots for the set up that I have tried with my schedule and also the report that I got. And as I said that everytime after I run the schedule my Test Id's that I used gets stuck :(

     

    Any help is greatly appreciated

    Your schedule design is close but needs a tweak or two

    1. Add an infinite loop for each User Group and then inside that infinite loop add the test

    The test won't run forever because it's time controlled by the Stage Duration as specified in the Schedule Element Details

    2. Tweak the change rate to add 1 user every 1/10th, 1/2, or whatever seconds

    3. I strongly suspect the 50 user run failed because some manual correlation is needed, this is common.  A common example is a JSESSIONID being sent as part of the request header or a cookie back to the server.  This has to be unique for every user ID.

    If you just play back the same test with different user logins and they send the same JSESSIONID to the server, the server will think the request is invalid.  So the test needs to be modified, in my example, not sure yet for yours, to use an unique JSESSIONID for each login.

    Try this:

    1. after the test finishes right mouse click on the Performance Test run in the Test Navigator and choose Display Test log

    2. click on the Events tab next to Overview and Data Correlation

    3. highlight the name of the schedule and right mouse click and do Expand

    4. click on any page request

    5. on the Protocol data tab click on Event Log

    This will let you know what's going on in the test

    See attached

     

     

     

    Attachments

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-14T17:14:07Z  

    Your schedule design is close but needs a tweak or two

    1. Add an infinite loop for each User Group and then inside that infinite loop add the test

    The test won't run forever because it's time controlled by the Stage Duration as specified in the Schedule Element Details

    2. Tweak the change rate to add 1 user every 1/10th, 1/2, or whatever seconds

    3. I strongly suspect the 50 user run failed because some manual correlation is needed, this is common.  A common example is a JSESSIONID being sent as part of the request header or a cookie back to the server.  This has to be unique for every user ID.

    If you just play back the same test with different user logins and they send the same JSESSIONID to the server, the server will think the request is invalid.  So the test needs to be modified, in my example, not sure yet for yours, to use an unique JSESSIONID for each login.

    Try this:

    1. after the test finishes right mouse click on the Performance Test run in the Test Navigator and choose Display Test log

    2. click on the Events tab next to Overview and Data Correlation

    3. highlight the name of the schedule and right mouse click and do Expand

    4. click on any page request

    5. on the Protocol data tab click on Event Log

    This will let you know what's going on in the test

    See attached

     

     

     

    Thanks again for your responses!

    1. I have updated my schedule as you mentioned, please see the attached screen shots of the schedule, user group & data pool.

    I am not sure how to check or set  unique JSESSIONID for each login. I have 10 different log in ID's in the data table for both of the tests that I am using, is that okay?

     

    Regarding SSL :

    The URL that I hit usually don't get certificate error when I log in Normally, but when I record with RPT, I am getting certification error, not sure why it is only coming when recording with RPT.

     

    If you think the schedule set up is okay , then I will go-ahead and run. thanks for everything.

     

    Thanks

     

    Attachments

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-14T19:19:18Z  
    • RFTworker
    • ‏2013-11-14T17:14:07Z

    Thanks again for your responses!

    1. I have updated my schedule as you mentioned, please see the attached screen shots of the schedule, user group & data pool.

    I am not sure how to check or set  unique JSESSIONID for each login. I have 10 different log in ID's in the data table for both of the tests that I am using, is that okay?

     

    Regarding SSL :

    The URL that I hit usually don't get certificate error when I log in Normally, but when I record with RPT, I am getting certification error, not sure why it is only coming when recording with RPT.

     

    If you think the schedule set up is okay , then I will go-ahead and run. thanks for everything.

     

    Thanks

     

    The schedules look OK, you might experiment and make the change rate a bit faster than 2 seonds, as it'll take 40 seconds in the last stage get a full load, but it's not an absolute change rate, just something to experiment between runs

    Certificate errors

    Assuming  you're recording with IE, there's two steps to avoid the certificate error when recording

    It can be installed manually at

    It will be in the IBMIMShared directory, under plugins
    Look for a directory with a name like com.ibm.rational.test.lt.recorder.proxy_8.2.2.v20111130_1728\SSLCertificate
    The version # (8.2.2....) will be different
    in the SSLCertificate directory, you'll find RPTProxyCertificate.cer.

    Make sure it's installed to the Trusted Root Certification Auhorities

    Technote on this: http://www-01.ibm.com/support/docview.wss?uid=swg21372997

    <see attached picture1.jpg>

    Uncheck "Warn about certificate address mismatch"

    <see attached picture2.jpg>

     

    Updated on 2013-11-14T19:24:33Z at 2013-11-14T19:24:33Z by PaulLiskay
  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-14T19:23:43Z  

    The schedules look OK, you might experiment and make the change rate a bit faster than 2 seonds, as it'll take 40 seconds in the last stage get a full load, but it's not an absolute change rate, just something to experiment between runs

    Certificate errors

    Assuming  you're recording with IE, there's two steps to avoid the certificate error when recording

    It can be installed manually at

    It will be in the IBMIMShared directory, under plugins
    Look for a directory with a name like com.ibm.rational.test.lt.recorder.proxy_8.2.2.v20111130_1728\SSLCertificate
    The version # (8.2.2....) will be different
    in the SSLCertificate directory, you'll find RPTProxyCertificate.cer.

    Make sure it's installed to the Trusted Root Certification Auhorities

    Technote on this: http://www-01.ibm.com/support/docview.wss?uid=swg21372997

    <see attached picture1.jpg>

    Uncheck "Warn about certificate address mismatch"

    <see attached picture2.jpg>

     

    Is  the other set up for Schedule and log in ID's okay? I will now work on the certificate issue as per the above steps.

     

    Thanks

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-14T20:41:32Z  
    • RFTworker
    • ‏2013-11-14T19:23:43Z

    Is  the other set up for Schedule and log in ID's okay? I will now work on the certificate issue as per the above steps.

     

    Thanks

    Thanks for the Certification issue resolution. I have followed the steps and now I did not get the certification error.

    I just recorded a simple test, and ran it for 1 user and it ran fine.  I ran for 50 users ....When ever I run for 50 users, my ID get stuck and I am restarting my server again in order to log in back, what could be the issue with this , please help me.

    I see the below errors when this happens in the Event console

    General un-handled exception occurred during socket I/O read from web-server 'txeistestdc.txeis.net' retrieving URI '/Attendance/dwr/engine.js'.  Explanation message: 'java.io.IOException: An existing connection was forcibly closed by the remote host.'

     

    Please let me know if you want more details on the issue. Thank you!

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-15T02:27:51Z  
    • RFTworker
    • ‏2013-11-14T20:41:32Z

    Thanks for the Certification issue resolution. I have followed the steps and now I did not get the certification error.

    I just recorded a simple test, and ran it for 1 user and it ran fine.  I ran for 50 users ....When ever I run for 50 users, my ID get stuck and I am restarting my server again in order to log in back, what could be the issue with this , please help me.

    I see the below errors when this happens in the Event console

    General un-handled exception occurred during socket I/O read from web-server 'txeistestdc.txeis.net' retrieving URI '/Attendance/dwr/engine.js'.  Explanation message: 'java.io.IOException: An existing connection was forcibly closed by the remote host.'

     

    Please let me know if you want more details on the issue. Thank you!

    Hi at this point I strongly suspect you have data correlation issue.  Refer to the picture rpt2.jpg I posted above.  Get to that same layout in RPT. 

    In that I post I showed how to use the Test Log to show events on the Event Log under the Protocol data tab.

    Now you'll need to use the Event Log under the Protocol data tab to navigate to the failing command in the Test Log.

    Simply double click on the first yellow or red icon int he Event Log that's under the Protocol data tab.  This will pop up a dialog.

    Highlight the first message in this dialog that pops up then there's  a "Go to" pull down menu.  Choose Go to -> test log, then look at the failing request, request headers, response headers, response content by clicking on those tabs in the Protocol data pane.

    Chances are in the failing request is something that is a static value that needs to be made dynamic. This is called data correlation.

    While in the test log with the failing request still highlighted, then right mouse click and chose Compare with Test -> All and examine the differences.

    See attached screenshots for reference.

    This is the first part of learning data correlation: that is identifying what went wrong and what needs to be correlated.

    See if you can get this far with the tool.  It takes some practice, but the steps of displaying the test log, looking at the Event Log in the Protocol Data view, then navigating to the failed request, then comparing with the recorded step is a basic technique that needs to be learned in order to use RPT successfully

     

     

    Attachments

    Updated on 2013-11-15T02:28:17Z at 2013-11-15T02:28:17Z by PaulLiskay
  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-15T15:49:07Z  

    Hi at this point I strongly suspect you have data correlation issue.  Refer to the picture rpt2.jpg I posted above.  Get to that same layout in RPT. 

    In that I post I showed how to use the Test Log to show events on the Event Log under the Protocol data tab.

    Now you'll need to use the Event Log under the Protocol data tab to navigate to the failing command in the Test Log.

    Simply double click on the first yellow or red icon int he Event Log that's under the Protocol data tab.  This will pop up a dialog.

    Highlight the first message in this dialog that pops up then there's  a "Go to" pull down menu.  Choose Go to -> test log, then look at the failing request, request headers, response headers, response content by clicking on those tabs in the Protocol data pane.

    Chances are in the failing request is something that is a static value that needs to be made dynamic. This is called data correlation.

    While in the test log with the failing request still highlighted, then right mouse click and chose Compare with Test -> All and examine the differences.

    See attached screenshots for reference.

    This is the first part of learning data correlation: that is identifying what went wrong and what needs to be correlated.

    See if you can get this far with the tool.  It takes some practice, but the steps of displaying the test log, looking at the Event Log in the Protocol Data view, then navigating to the failed request, then comparing with the recorded step is a basic technique that needs to be learned in order to use RPT successfully

     

     

    Thanks for all your help!

    From your previous response....

    . I strongly suspect the 50 user run failed because some manual correlation is needed, this is common.  A common example is a JSESSIONID being sent as part of the request header or a cookie back to the server.  This has to be unique for every user ID.

    If you just play back the same test with different user logins and they send the same JSESSIONID to the server, the server will think the request is invalid.  So the test needs to be modified, in my example, not sure yet for yours, to use an unique JSESSIONID for each login.

     

    Can you please explain me how I can manually correlate JSession ID?

    I see the JSession ID as part of the URL and it is highlighted with Pink(I guess as reference). (please see the attached jession.doc)

     

    Also I see some errors in the event log for my 50 users schedule. I have followed above steps to navigate to Test log and compare. The errors show from the 2nd user onwards. please see the attached screenshot.(comparetestlog.doc).

    Some of them say "The request does not have a HTTP response status line" Not sure what this means.

     

    Your help is very much appreciated and I need this help for my December release.

    thank you so much!

     

     

    Attachments

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-15T16:26:23Z  
    • RFTworker
    • ‏2013-11-15T15:49:07Z

    Thanks for all your help!

    From your previous response....

    . I strongly suspect the 50 user run failed because some manual correlation is needed, this is common.  A common example is a JSESSIONID being sent as part of the request header or a cookie back to the server.  This has to be unique for every user ID.

    If you just play back the same test with different user logins and they send the same JSESSIONID to the server, the server will think the request is invalid.  So the test needs to be modified, in my example, not sure yet for yours, to use an unique JSESSIONID for each login.

     

    Can you please explain me how I can manually correlate JSession ID?

    I see the JSession ID as part of the URL and it is highlighted with Pink(I guess as reference). (please see the attached jession.doc)

     

    Also I see some errors in the event log for my 50 users schedule. I have followed above steps to navigate to Test log and compare. The errors show from the 2nd user onwards. please see the attached screenshot.(comparetestlog.doc).

    Some of them say "The request does not have a HTTP response status line" Not sure what this means.

     

    Your help is very much appreciated and I need this help for my December release.

    thank you so much!

     

     

    I just observed in the test log  that RPT is using the same JsessionID , for all the user ID's

     

    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0
    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0
    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0

     

    Thanks

     

  • PaulLiskay
    PaulLiskay
    644 Posts

    Re: More time on Page Performance report

    ‏2013-11-15T18:55:59Z  
    • RFTworker
    • ‏2013-11-15T16:26:23Z

    I just observed in the test log  that RPT is using the same JsessionID , for all the user ID's

     

    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0
    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0
    JSESSIONID=F4A8DB7E52381AF6762BD18A04DAFCC0

     

    Thanks

     

    Hi, proof that the JSESSIONID is dynamic would be if you do another recording or look at a previous recording and look for the JSESSIONID, they're probably different.

    A little learning is going to be necessary, please refer to the 8.5 document on data correlation here http://pic.dhe.ibm.com/infocenter/rpthelp/v8r5m0/index.jsp?topic=%2Fcom.ibm.rational.test.lt.common.doc%2Ftopics%2Ftdatacorrtop.html

    The second topic "Data Correlation Overview" has a video that will show how to do corelation

    See attached screenshot

    Basic skills to learn are:

    1. using the Test Search for a potential reference to be used later on in the test, in your example copy "F4A8DB7E52381AF6762BD18A04DAFCC0" the clip board, then right mouse on the name of the of the test in the Test Editor and do Test Search

    Search HTTP Responses for that value.  Find the first instance of it, then highlight it and make a refernce out of it.

    Once a reference is created it'll ask to find possible substitions.  Use the dialog to find matches and then subsstitute it, save the test and play it back

    If the test fails, then go through the cycle of displaying the test log, bring up event log in the Protocol data tab, then click on the first yellow or red message, goto test log, etc.

    The first basic concept to understand is ALL references must come from a previous Response or Response header.

    ALL substitutions from references are made in requests, request headers, or request data.

    Also, do not try to correlation fields in a cookie that RPT identifies as a "Server Supplied Cookie", it's a waste of time

     

     

     

     

     

     

     

    Attachments

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-19T15:04:38Z  

    Hi, proof that the JSESSIONID is dynamic would be if you do another recording or look at a previous recording and look for the JSESSIONID, they're probably different.

    A little learning is going to be necessary, please refer to the 8.5 document on data correlation here http://pic.dhe.ibm.com/infocenter/rpthelp/v8r5m0/index.jsp?topic=%2Fcom.ibm.rational.test.lt.common.doc%2Ftopics%2Ftdatacorrtop.html

    The second topic "Data Correlation Overview" has a video that will show how to do corelation

    See attached screenshot

    Basic skills to learn are:

    1. using the Test Search for a potential reference to be used later on in the test, in your example copy "F4A8DB7E52381AF6762BD18A04DAFCC0" the clip board, then right mouse on the name of the of the test in the Test Editor and do Test Search

    Search HTTP Responses for that value.  Find the first instance of it, then highlight it and make a refernce out of it.

    Once a reference is created it'll ask to find possible substitions.  Use the dialog to find matches and then subsstitute it, save the test and play it back

    If the test fails, then go through the cycle of displaying the test log, bring up event log in the Protocol data tab, then click on the first yellow or red message, goto test log, etc.

    The first basic concept to understand is ALL references must come from a previous Response or Response header.

    ALL substitutions from references are made in requests, request headers, or request data.

    Also, do not try to correlation fields in a cookie that RPT identifies as a "Server Supplied Cookie", it's a waste of time

     

     

     

     

     

     

     

    Thanks for your notes and suggestions. they are really helping me.

    I just recorded a new test with one user and played back and saw that the first page was showing 14 seconds which is not correct.

    Also I did not see any data correlation errors, or any warnings or any errors in the error log.

    any ideas?

  • RFTworker
    RFTworker
    13 Posts

    Re: More time on Page Performance report

    ‏2013-11-20T17:42:44Z  
    • RFTworker
    • ‏2013-11-19T15:04:38Z

    Thanks for your notes and suggestions. they are really helping me.

    I just recorded a new test with one user and played back and saw that the first page was showing 14 seconds which is not correct.

    Also I did not see any data correlation errors, or any warnings or any errors in the error log.

    any ideas?

    I have contacted IBM on this issue and it got resolved. We followed some tips from IBM for setting up HTTP tests and schedules.  Some of them are below:

    1. Rate of change is set to one user for every 3 seconds, instead of all users over
    2. Statistics sample interval is set to 5 seconds, instead of 15 seconds to retrieve the data from the server
    3. Whenever the test play starts automatically selected to clear cookie
    4. Cleared page cache when recording and when test starts
    5. Disabled page cache emulation in tests
    6. Forced the garbage collection manually to know the size the memory for the host controller machine.

    Thank you very much for your support and continuous response!! Very much Appreciated

     

    Thanks