Java memory crash when generating report (Gatling 2.0)

Gatling (Open-Source)
1

Hi All,

I have a scénario.

The run is OK but i have a java crash memory during the report generation.
So, the report is not generated and a dump file java_pid3036.hprof is present

In attached my simulation.log that causes the crash.
I don’t attach the .hprof in group because 534 Mo. Tell me if we need it.

I tested several versions of gatling 2.0. The crash is also present on yesterday snapshot.

Thanks for you help.

Cheers,

Franck

simulation.zip (185 KB)

2

Here a link to a biggest simulation.log of another run for the same scenario with more users :
https://drive.google.com/file/d/0B4QSMpXM2ZENSjFuNEk1ZkZyZVE/edit?usp=sharing

3

https://github.com/excilys/gatling/issues/2030

I’m working on the fix, stay tuned.

4

But then, the root of the problem is that you have ~14.000 different request names!!!
Gatling reports will be pretty useless then.

5

Fixed

6

I love to push and also stress Gatling in their limits !! :wink:

Thanks for the fix.

I’ll test soon the last snapshot that day.

Franck

7

I tested the generation :

  • for the first simulation.log with 14.000 request names => It’s OK.

  • for the second simulation with 254.000 request names => the report generation is always KO for memory.

I have others runs with more users than the second one.

So, i realize now that this huge number of request names.

I will change my strategy to reduce the request name number that becomes not acceptable.

Franck

8

Gatling reports were definitively not designed for this.
Could you elaborate about your use case, please?