DevOps Automation

 View Only
Expand all | Collapse all

What argument should be passed to to close connection after every request

  • 1.  What argument should be passed to to close connection after every request

    Posted Sun May 30, 2021 09:05 PM

    Hi I found this online "DKEEP_ALIVE_ACROSS_LOOPS_WITHIN_TEST=false" to close connection during load test instead of Keep-Alive, but seems like this ARG is not working properly. Can you please help ?


    #Support
    #SupportMigration
    #RationalPerformanceTester


  • 2.  RE: What argument should be passed to to close connection after every request

    Posted Tue June 01, 2021 04:14 PM

    how are you using this argument, the default value is false so not using it and using it with false should have the same behavior

    this argument can only be used as a VM_ARG for a load agent which means the schedule must be using an agent not running on local computer

    What specific steps have been done in using this argument and what behavior is seen that makes it look like it's not working?

    Thank you


    #SupportMigration
    #Support
    #RationalPerformanceTester


  • 3.  RE: What argument should be passed to to close connection after every request

    Posted Tue June 01, 2021 04:56 PM

    I am passing this argument in agent Property using RPT_VMARGS. And yes i am using agent while running test. While using this argument, we are still seeing Connection:Keep-Alive in response header.


    #RationalPerformanceTester
    #Support
    #SupportMigration


  • 4.  RE: What argument should be passed to to close connection after every request

    Posted Tue June 01, 2021 05:48 PM

    OK, take the default value of false which is also default behavior, the keep-alive is coming from the server which RPT has no control over. RPT will honor the keep-alive request if possible.

    When the test is finished, RPT will close the connection, the test is over, no need to keep it open.

    What keep alive = true means is that even when the test is over, RPT will *not* close the connection so the next time the test is called from the loop in the scheudle, it'll reuse the opened connections from the previous test.

    This can save possibly running out of "handles" or TCPIP ports

    This is in the doc here https://help.blueproddoc.com/rationaltest/rationalperformancetester/10.1/com.ibm.rational.test.lt.doc/topics/taddloops.html

    "Loops in schedules close the server connection at the end of each iteration and reopen it at the beginning of the next iteration. This action models the behavior of a user closing and reopening a browser."

    So keep alive across test just tells RPT to not close the server connection at the end of each iteration and reopen it at the beginning of the next iteration. You won't see any difference in the Response header for "Keep-alive" at all


    #SupportMigration
    #Support
    #RationalPerformanceTester