Artillery: Bug: arrivalRate and arrivalCount in config are not respected in test run

[ericmacfarland]

Versions

Node: v8.1.3
Artillery: v1.6.0-2

Backround

I have a .yml script that has the intent of running a high volume burst of requests. (See example-file.txt)
Details:

• One phase with duration: 1 and arrivalCount: 340
• Eight scenarios, each with one request

Expected result

When the test is run, 340 scenarios should be fired

Actual result

When the test is run, 500 scenarios are fired, final log below:

Complete report @ 2017-07-12T14:43:03.286Z
  Scenarios launched:  500
  Scenarios completed: 500
  Requests completed:  500
  RPS sent: 8.89
  Request latency:
    min: 42.8
    max: 54254.8
    median: 6674.8
    p95: 52835.3
    p99: 53474.8
  Scenario duration:
    min: 51.3
    max: 54266.4
    median: 6695.5
    p95: 52840
    p99: 53490
  Scenario counts:
    /api/user-settings/featureToggles: 76 (15.2%)
    /dashboard: 60 (12%)
    /api/dealer/dashboardCards: 81 (16.2%)
    /api/user-settings/profile: 54 (10.8%)
    /api/dealer/stepsToGrow: 60 (12%)
    /api/user-settings/users/{{ email }}: 57 (11.4%)
    /api/dealer/ensembleView: 54 (10.8%)
    /api/user-settings/ensembles?userId={{ user }}: 58 (11.6%)
  Codes:
    200: 500

The same thing happens when arrivalCount: 240, but the actual count is 250 scenarios fired.

[hassy]

Interesting, thanks for the bug report @ericmacfarland. What version of Artillery & Node.js is this?

[ericmacfarland]

@hassy I've updated my first post with my versions

[hassy]

Thanks. I'm able to reproduce the bug.

[ericmacfarland]

Thanks for the quick replies!

[hassy]

The bug's been fixed, the fix will be available in the next release of Artillery (to be published this week).

[ericmacfarland]

Fantastic, thanks!