When an error occurs, a SampleResult with 0 timestamp is generated

[Former user]

During a test we face some errors on server side, this is what gets generated:

0,0,WS-Ping-Pong,Sampler error,"Sampler must use existing connection, but there is no connection",UTGP-i-0d4d88f4bab08c111 1-39,,false,,0,0,231,462,null,0,0,0,XXXXX,4800000,1066,1066,51006,2.54,0.0,49384,0

As you can see, the first column is the timestamp and it is equal to 0, which breaks the JMeter report for example.

[PhilippeM]

Hello,

I created this issue for information.
Regards

[PhilippeM]

PR:

• https://bitbucket.org/pjtr/jmeter-websocket-samplers/pull-requests/7/fix-issue-196-by-ensuring-samplestart-and

‌

[Peter Doornbosch]

Hi Philippe,

Did you see this one: https://bitbucket.org/pjtr/jmeter-websocket-samplers/issues/188/sample-start-time-is-incorrect-when-there?

If i’m interpreting your PR correctly, your fix will lead reporting tools to believe the sample took 0 milliseconds, which is not correct, because no request was sent at all. Maybe the problem is that JMeter does not take into account that an sample without a result (but only with an error) is also a possibility.

Cheers,
Peter

[PhilippeM]

Well technically, the sampler started and failed.
I need to double check this, but if I remember correctly, in JMeter the failures are excluded from response time computing in the HTML Report.

sampleStart initializes the timestamp at which sampler triggered.

My fix tries to avoid having sampler without start time set.

I saw the other issue, I would suggest that you report it as a bug if you think it is.
Thanks for your rapid feedback anyway.

[PhilippeM]

Hello @Peter Doornbosch ,

Do you have any update ?
Thanks