Active users compressed in long-duration reports

I’m seeing something strange with number of active users in my reports.

I see a graph of active users that looks to me like the active users have been compressed into a narrower time frame than the actual test duration. For example, my scenario is to ramp from 0 to 20 users per second over 1 minute, then hold for 30 minutes, then ramp back down over 1 minute. The active users graph seems to match that, but it does it in only 15 minutes, and then reports 0 for the rest of the duration.

However, I changed the duration to 5 minutes of sustained load, and the active users graph looks as one would expect. However, strangely, when I was running very long duration tests (16+ hours), the active users appeared to be the right shape, but compressed to the left by a wide margin. I don’t remember exactly by how much, but I don’t know if it was 15 minutes or something longer. Sadly, I accidentally deleted those reports, so I can’t confirm. I will run another long duration scenario tonight and attach a simulation.log file later.

Is this a bug?

I started with the gatling-sbt-plugin-demo, and the only config tweak I made was to change the indicator category thresholds to 500ms and 1000ms. My SBT config has me loading Gatling 2.2.3.

Confirmed: I just looked at a 6-hour run. The Active Users graph is compressed into about the first 15 minutes of the report. Looks like a reporting issue.

The resulting simulation.log, even compressed (7MB), is too big to attach. If you need to see it, let me know the best way to get it to you.

This is a definite bug in the report generator code of Gatling 2.2.3. The Active Users portion of each graph takes up a fraction of the total duration, instead of being stretched out over the full duration as it should be. It only happens with runs over a certain length, though I have not narrowed down exactly what duration that is. 30 minute runs are enough to see it. The longer the run, the more pronounced the effect.

I first reported this a month ago, but there was no response, so I don’t know if anyone saw it.

If you prefer me to submit a defect directly to GitHub, let me know.

Hi John,

Have you checked latest releases and their release notes (this mailing list, Twitter or Github)?

I had not. But I see upon checking that it is fixed in the 2.2.5 release. Excellent. Thank you!

As a side note, when I went to GitHub and looked for the release notes, I couldn’t find them. When I click on Releases (, it only shows up to 2.2.0. Am I doing something wrong?

Also, for my future reference, do you prefer me to communicate potential bugs here first, or go straight to GitHub?