| Summary: | JMeter Statistical mode in distributed testing shows wrong response time | ||
|---|---|---|---|
| Product: | JMeter - Now in Github | Reporter: | Eagle Liu <liu.xp2003> |
| Component: | Main | Assignee: | JMeter issues mailing list <issues> |
| Status: | RESOLVED FIXED | ||
| Severity: | regression | CC: | p.mouawad |
| Priority: | P2 | ||
| Version: | Nightly (Please specify date) | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | All | ||
| Bug Depends on: | 52125 | ||
| Bug Blocks: | |||
| Attachments: |
JAVA sample script to reproduce the problem
JMeter log for Bug 52339 |
||
|
Description
Eagle Liu
2011-12-15 09:17:59 UTC
Is it only statistical mode that has the problem? Does the same problem occur if you replace the HTTP samplers with Java samplers? This will eliminate the effects of accessing the server from different hosts. I have verified with Standard,Batch,Hold and Statistical mode. It was only occurred on Statistical mode. I have justed checked with BeanShell Sampler. Both of HTTP and BeanShell sampler have this problem. Can you try with Java Sampler please? I cannot reproduce the problem. Created attachment 28075 [details]
JAVA sample script to reproduce the problem
The issue could be reproduced by JavaTest class. I attached script. It could help you reproduce the issue. I'm still not seeing the problem, but have not yet been able to test across two different physical hosts. Also I don't see anything obvious in the changes to StatisticalSampleSender since 2.5.1. I have recorded about the problem with video. And you can download from the URL "http://speedy.sh/RfBCj/JMeter-Bug52339.zip" to review it.[The file size is about 5.14MB. MD5: 43F7E75749DFFC778B7C53864271550A] I have made it to execution file. You can open it directly to replay it. Can you run the Java test sample you provided, and attach the following output: jmeter.log from client and server systems JTL file from the run (make sure subsamples are saved). Also, please do the same for the test run as non-GUI on the server node. Obviously in that case cannot use Statistical mode. Created attachment 28082 [details] JMeter log for Bug 52339 Run_C-S_clientLog.zip and Run_C-S_serverLog.zip is the first run log. Run_no-GUI_serverlog.zip is the second run log. The problem is not happened when running as non-GUI on the server node. jmeter.properties has not been changed on server node. run command: jmeter -n -t Sample_JMeterBug_52339.jmx -l runNOGUI-onJMeterServer.jtl Thanks for the JTL files - I think I've found the problem. Try using Summary Report instead of Aggregate Report, and the figures look OK. Seems that Aggregate report is not allowing for the fact that Statistical results represent multiple samples. I'll try and post a fix for that shortly. Looks like the problem might have been caused by the fix for Bug 52125. In the meantime, use Summary Report rather than Aggregate Report and the display problem should go away. Please can you confirm if that fixes the issue? Revision 1197375 is ok for Aggregate Report. Summary Report is correct in apache-jmeter-r1214179 build. Aggregate Report is incorrect in apache-jmeter-r1214179 build. Also fixes Bug 52125 URL: http://svn.apache.org/viewvc?rev=1221486&view=rev Log: Bug 52339 - JMeter Statistical mode in distributed testing shows wrong response time< Modified: jmeter/trunk/src/jorphan/org/apache/jorphan/math/StatCalculator.java jmeter/trunk/src/jorphan/org/apache/jorphan/math/StatCalculatorInteger.java jmeter/trunk/src/jorphan/org/apache/jorphan/math/StatCalculatorLong.java jmeter/trunk/test/src/org/apache/jorphan/math/TestStatCalculator.java jmeter/trunk/xdocs/changes.xml This issue has been migrated to GitHub: https://github.com/apache/jmeter/issues/2681 |