Configuring ramp up in nGrinder

[ivannov]

Hello,

I would like to run a load test with nGrinder with the following configuration:

Total number of vusers: 1000
Total number of agents: 1
Number of threads: 200
Number of processes: 5

Besides that I would like to have a smooth ramp up of my users. That is why I enabled Ramp Up and set incremental step to 4 and interval to 10,000 ms. This way I expect that after 500 seconds (8:20 minutes) all of the vusers will be running. Here is the configuration screen:



However, now the test is running already for more than 40 minutes and the UI reports that I have 618 vusers running in total and only 4 processes:



Did I set anything wrong?

I just want to mention that my test script simulates a 10 seconds think time before every request by this code in the script:

    @Before
    public void setup() {
        grinder.sleep(10000)
    }

Is this the right way to do that?

Thanks and regards,
Ivan

[junoyoon]

Your test seems to be very heavy (such as much memory consumption...).

Usually this happens the when test threads are failed during start up.

Could you decrease the vuser size to 500 and add one more agent?

2015-07-16 21:16 GMT+09:00 ivannov [via ngrinder] <[hidden email]>:

Hello,

I would like to run a load test with nGrinder with the following configuration:

Total number of vusers: 1000
Total number of agents: 1
Number of threads: 200
Number of processes: 5

Besides that I would like to have a smooth ramp up of my users. That is why I enabled Ramp Up and set incremental step to 4 and interval to 10,000 ms. This way I expect that after 500 seconds (8:20 minutes) all of the vusers will be running. Here is the configuration screen:



However, now the test is running already for more than 40 minutes and the UI reports that I have 618 vusers running in total and only 4 processes:



Did I set anything wrong?

I just want to mention that my test script simulates a 10 seconds think time before every request by this code in the script:

    @Before
    public void setup() {
        grinder.sleep(10000)
    }

Is this the right way to do that?

Thanks and regards,
Ivan

---

If you reply to this email, your message will be added to the discussion below:

http://ngrinder.642.n7.nabble.com/Configuring-ramp-up-in-nGrinder-tp1823.html

To start a new topic under ngrinder-user-en, email [hidden email]
To unsubscribe from ngrinder-user-en, click here.
NAML

[ivannov]

Thanks! I think I found the sweet spot: two VMs, running one agent each, 5 processes per agent and 100 threads per process. Thus I achieved 1000 vusers.

Thanks a lot for the quick help!