Session graph is not constant


I’m new to Gatling but I like it already especally for the ease of setting up scenarii and the exploitation of the results.

My problem is when I user a ramp in a scenario, the session number per second should be constant (except at start and end) all along the test.
But the curve presents peaks every 35/40 seconds where the number of sessions by second is three to four times the cruise value (which i expect).

My scenario is relatively simple (two requests, one local and one distant) and here is the code to manage the users :

What am I doing wrong?

If needed I can post a report to see what’s going on…

Thanks in advance for the help.


Mmmm, I'm wondering if you might be bitten by this Akka issue:

Could you try to change the ramp duration to something that's not a
multiple of 50 (which is the tick duration set up in application.conf),

Here is the gist of the scenario...

really simple :

Thanks again...

Could you also provide your simulation.log file, please?

I see that your gist uses 2000 users and not 20000 as you first stated.
Is it that you lowered it? Does the problem still occur with 2000 users?

Lowering the number of users didn't solve the problem. The report
graph just look different, less peaks which is consistent with the

Here is a gist to the simulation.log file :

IMHO, this is perfectly normal.

Your scenario is so short (only 2 requests, and the first one looks that a static page, so response time should be almost immediate) that your users’ lifespan is directly correlated to the second request response time.

Have a look at ps.psp response time and you’ll realize that it periodically jump from ~150ms to 25s (GC on your server maybe), making your users stay alive much longer.



Thanks Stephane.

It was from the server. Nothing to do with gatling… oups :-/

No pro, that was an interesting case.