phases settings

[nnnikolay]

Hello,

Could you please explain a bit how that section is working? because from the documentation it's not clear for me.

I have the following config

  phases:
    - duration: 120
      arrivalRate: 1

Which I would understand as "within 2 minutes increase concurrency from 0 to 120 of virtual users"
is that means that on the 5th second we would have 5 virtual users in total or 5 virtual users will be added to current amount? Which means on the 5th second we would have already 15th virtual users.

Here is the output of the last two reports

Report @ 10:36:41(+0200) 2017-09-25
  Scenarios launched:  1
  Scenarios completed: 5
  Requests completed:  19
  Concurrent users:    4
  RPS sent: 3.76
  Request latency:
    min: 84.1
    max: 1665
    median: 657.1
    p95: 1619.5
    p99: NaN
  Scenario duration:
    min: 3187.8
    max: 3927
    median: 3813.4
    p95: NaN
    p99: NaN
  Codes:
    200: 13
    201: 6

All virtual users finished
Summary report @ 10:36:41(+0200) 2017-09-25
  Scenarios launched:  120
  Scenarios completed: 120
  Requests completed:  600
  Concurrent users:    635
  RPS sent: 4.9
  Request latency:
    min: 83.1
    max: 3764.2
    median: 773.9
    p95: 2307.9
    p99: 3013.8
  Scenario duration:
    min: 3187.8
    max: 8851.6
    median: 4716.2
    p95: 7373.7
    p99: 8785.9
  Scenario counts:
    order request: 120 (100%)
  Codes:
    200: 360
    201: 240

I mean for me is quite hard to make a conclusion in the report (apart from that the requests are slow by them self obviously).

I've only one scenario which consists from 5 http reqs.
I see that 120 scenarios completed, but what is confusing me is a number of parallel users .. 635, really? And these guys were able to finish only 120 scenarios?

Looks like either I don't understand something or misconfigured, I would highly appreciate your help in understanding this.

Thanks

[hassy]

635 concurrent users being reported is a bug. Thanks for reporting that @nnnikolay.

  phases:
    - duration: 120
      arrivalRate: 1

The meaning of this phase is "spawn 1 virtual user every second for 120 seconds". The number of concurrent users on second N depends on what the users do (the scenario definition). For example: if each user only makes 1 HTTP request which is served very quickly, you may not have more than 1 concurrent users at any given point. However, if you have a think: 200 in your scenario definition, then once the load phase ends, you will have 120 concurrent users.

Does that make sense?

[nnnikolay]

Frankly, I was not able to find out think: definition in the documentation. Therefore I'm not sure that I understand why we will end up with 120 concurrent users. Could you please point me to the doc where it says what is that?

with arrivalRate is clear now, thanks.

One more question:

Report @ 12:00:06(+0200) 2017-09-25 
...
All virtual users finished
Summary report @ 12:03:16(+0200) 2017-09-25

I was expecting to have a load test for a 60 sec with the following settings:

    - duration: 60
      arrivalRate: 3

Why did it take 3 minutes? Because it was waiting when for all the requests are responded?

And another question is: which report's value should I use as an orienteer RPS sent? I mean what is the best practice in your opinion.

And last one: how could I know how many (in total) users has passed the scenario (Scenarios completed?)?

In the final report I see following numbers:

Summary report @ 12:03:16(+0200) 2017-09-25
  Scenarios launched:  180
  Scenarios completed: 50
  Requests completed:  550
  Concurrent users:    17469
  RPS sent: 2.76
  Request latency:
    min: 142.8
    max: 60234.1
    median: 30216.9
    p95: 59884
    p99: 60205.5
  Scenario duration:
    min: 10281
    max: 175389.5
    median: 137088.8
    p95: 174195.5
    p99: 175389.5
  Scenario counts:
    order request: 180 (100%)
  Codes:
    200: 204
    201: 185
    504: 161

I see that there were 180 scenarios launched but only 50 is completed, means successfully or in general? And if in general what has happened with others?

btw, just in case

artillery --version
1.6.0-9

Sorry, for so many questions :-)