0
votes

I am using Apache JMeter 2.13 on Windows 7 and Java 1.8.0_73

I have a JMeter file set up with a thread group with 1 thread, 0 ramp-up period and 5 loop count. I've recorded the log in process for a web application using the recording controller, the HTTP(S) test script recorder and Firefox. I have added the other elements you can see in the screenshot below

JMeter script setup

I've then parameterized several of the requests to extract login tokens and moved the root URLs out into user variables (so that I can run it against my local build or any of the server builds) including a beanshell script for dynamic authentication headers in some of the requests.

When I clear results and then run the test script I see that the first request in the first iteration has a sample time of around 10 seconds, the same is true of the 4th request (the next major page). The second iteration these same request takes ~600ms, the third iteration 350 e.g.

10318, 605, 350, 641, 310

I then clear results, run again with 10 loops

10366, 1012, 600, 300, 293, 632, 324,292, 331, 291

So my questions are: Is this expected JMeter behaviour? Is there anyway to change it

1

1 Answers

0
votes

Usually you have to have some warmup method to be run first. Just to set up all the stuff, to reach the db, start some threads, etc.

The results of the warmup are not taken into account when analysing the performance.