load test response time

Hi Larry,

I'm sorry, but the attached image doesn't show a lot of information needed to assist here. It'd be great if we could see:
- number of threads in your strategy
- focus which steps are "step 1", 2, 3, etc. At the moment there are additional half cropped disabled and other names steps that could refer to your question

Please provide the additional information so that we could assist you further. Thanks,
Michael Giller
SmartBear Software

Michael,
Thanks for the responding. Below are my answers to your questions.
- number of threads in your strategy
A: 100
- focus which steps are "step 1", 2, 3, etc. At the moment there are additional half cropped disabled and other names steps that could refer to your question
A: "Step 1", 2, 3 are the same function (sendMsg and read respMsg in Q) except sleepTime 20 sec. All the disabled test steps are ignored since they are not included in the running scenario.
My question: Q_1 (step 1) min (2986 ms) and Q_2 (step 2) min (2748 ms) load test result data were not logic since there were sleep time in both test steps (or scripts: Q_1: 20 sec; Q_2: 40 sec)
Thanks,
Larry

Hi,

I would suggest you try with 10 thread and 1 run and see if the results are correct.
If you look at the average time and max time the values look OK, the minimum time could be due to the number of threads being used. I can have development take a look at this if the results appear to be skewed still.



Regards,
Marcus
SmartBear Support

Marcus,
I tried 10-thread with 1 Runs per Thread scenario and the min data looks fine.
Thanks,
Larry

Hi,

So the minimum appears less than what you expect with a high number of threads such as 100? Or are you able to run with a 100 threads again and get the minimum you would expect? It's possible that because of the high number of threads that if all you are doing is sleeping in the scripts that they could finish faster as the sleep method does not guarantee the amount of time the script will be idle. If you are still having the minimum be much lower than you would think with 100 threads or more I can have development take a further look at this. Also, there is a thread cap at 200 threads during a load test. If you would like to increase this amount then you will need to use the -Dsoapui.threadpool.max SoapUI jvm property. See this link for more information http://www.soapui.org/Load-Testing/threads.html.



Regards,
Marcus
SmartBear Support

Hi Marcus,
Yes, 100 threads reproduced the issue. My script is not only doing sleeping but also send/read JMS message to/from the Queue. "sleep method does not guarantee the amount of time the script will be idle" does not make sense to me. sleep(20000) and sleep (40000) should be different idle time, right?
Thanks a lot for the Threads setting valuable information link.
Larry

Hi,

My script is not only doing sleeping but also send/read JMS message to/from the Queue.

I will have to double check on this. Since your load testing using JMS this could potentially cause problems, especially with the 100 threads being used. Also, your doing this programmatically using Groovy.



Regards,
Marcus
SmartBear Support

Thanks Marcus. Please let me know if you need further information.
Larry