jtlWriter to only include samples matching regex in log, like influxDbListener

[DmitriyStroganov]

Hi there.

Thank you for you wonderful job on this DSL.

I think it can be useful to tune jtlWriter to only include samples matching regex, in cases when you place all the samplers that are doing the business step in transaction controller to have the duration of the step.

So you can filter out the samplers, but leave transactions in a specific log to have there a total times of how long does it take to perform the whole business step.

Now you can use filter results plugin for this, but I believe it should be good to have this functionality in the box.

Thanks for the attention, have a good day.

Dmitriy.

[andreig09]

Hi @strodi, thanks for reaching out.

We were discussing this suggestion and have some doubts about it.

Typically, you can use the transaction controller to measure the entire business step, as the response time for this element is saved in the results .jtl file. Does the approach you're suggesting add unique value in some other way or provide different information?

Additionally, you could filter the .jtl file using regex in Java code. While this isn't directly implemented in the jtlWriter, it can be done within the same script.

We're asking these questions because, by default, we aim to avoid adding extra functionality or complexity to components unless it's absolutely necessary. Our goal is to keep JMeterDSL as simple as possible.

Thanks,

[DmitriyStroganov]

Hi there!

Thanks for the reply.

No, I am not suggesting to add the additional info or add any unique values.
The idea is to have only transaction controllers (filtered by name using regex) in one of jtls that are written by jtlWriter to have the summary on the business level.
Now there is the option to record only errors or successes, but no option to write only specific samplers in jtl log, this is what I meant.

I didnt know that it is possible to filter .jtl in Java code.
I'll look in this direction, thanks.

Thank you and all the developers of this fantastic tool again.
The work you done is making jmeter (great and time proof solution) expertise alive against some modern things like gatling, locust or k6.

[andreig09]

Thanks again for your kind words, @strodi. It really means a lot to us that the community recognizes the value of this tool, uses it, and contributes, as you're doing with this discussion.

I believe I now understand what you were requesting. You can definitely achieve that with Java code in the script. Since JTL files are typically XML (or CSV), you can parse the original file, filter the samples by their label attribute using a regex, and write the filtered results to a new JTL file.

That said, I'll discuss with the team the possibility of adding this as a feature to jtlWriter, as it would make the configuration much more straightforward.

Thanks again!