The response times are normal within 2-3 second range when running in fiddler (while the performance test is ongoing) but the jmeter 5.2 results are >100 seconds for most of the transactions. I have 8 thread groups performing different scenarios and 5 threads for each group. Can anyone please suggest what could be wrong? Im using constant throughput timer to pace the execution for each thread so that the desired volumes are achieved (5/min). Will that be an issue? If so, what is the best option. Thanks you for any help. I need to get the tests done ASAP. Any help would be much appreciated.
I can think of 2 reasons:
- Given 8 Thread Groups with 5 virtual users you're accessing your application with 40 concurrent users and it might be the case the application gets overloaded hence response time increases as concurrency increases. Try running your test with 1 virtual user. If you get response times comparable to what you see in Fiddler than your system cannot properly handle even 40 users. Check its logs, configuraiton, baseline OS health metrics (CPU, RAM, Network, Disk, Swap, etc.), this can be done using JMeter PerfMon Plugin
If you're using Fiddler as a sniffer for the browser it might be the case that the browser is caching |heavy" content (images, scripts, styles, sounds, fonts, etc). Tick "Disable Cache" in your browser developer tools, it will trigger "clean" sessions (like JMeter does)
Going forward add HTTP Cache Manager to your test plan so JMeter would have its own browser cache implementation.