Spike test with 2 parallel scenarios

I d’ like to buid a spike test in open model with 2 parallel scenarios - 1st is normal load and 2d is a spike inside of the 1st.
What I cannot do is to rump down a spike where I want it.
Is it possible to do?


Well, yes, of course!
Have you gone through the tutorials? They demo concurrent scenarios: Gatling - Advanced Tutorial

I not only read documentation but a lot of threads about ramp down also. It seems that now you can only cut users by terminating a test or by default they ramp down with the speed of ramp up. This doesn’t work for me as I’d like to ramp down a spike long before test termination.

Not sure what you mean by spike, then. In Gatling, a spike is stressPeakUsers, see Gatling - Injection

In Gatling, you control when users get started, not when they terminate (they complete their scenario normally).
You can ramp down the users arrival rate, not the number of concurrent users.

Ramping down the number of concurrent users doesn’t make sense. If that’s what you’re trying to achieve, I recommend you have a look at the difference between open and closed models: Gatling - Injection

I also did a talk last week at Devoxx where I explain this: Load Testing Crash Course with Gatling by Stéphane Landelle - YouTube

In the production logs I can only see the number of concurrent users not the arrival rate. That’s what I’m trying to emulate. And by spike I mean a sudden burst in users number which can last for some time, so controlling a user’s number but not arrival rate makes sence to me (though we have an open model load).
May be I should count the arrival rate myself then.

Yes, sadly, that’s one of the 2 reasons people try to go with a closed model while their system isn’t: because they monitor a wrong metric in production (eg Users in Google Analytics).
The other reason is because they were trained this way on some commercial tools many years ago.