Carrying on data from a scenario to the next in k6 load test

Question

I have the following scenarios setup for a load test, and they are expected to run in series:

export let options = {
  scenarios: {
    companyScenario: {
      executor: 'per-vu-iterations',
      exec: 'runCompanyScenario',
      vus: 1,
      iterations: 1,
      maxDuration: '2s',
    },
    employeeScenario: {
      executor: 'per-vu-iterations',
      exec: 'runEmployeeScenario',
      vus: 1,
      iterations: 1,
      maxDuration: '10s',
      startTime: '2s',
    }
  },
};

I'm looking for a way to store the ID of a resource created in the first scenario ("companyScenario"), which is given as part of a response, and then use that ID as part of payload in the next scenario ("employeeScenario").

I tried to use a global constant approach, but the value is lost in between.

Anybody has a hint?

Answer 1

From your scenario definitions, it looks like you are (ab)using scenarios to implement a test lifecycle with a setup phase. k6 already supports a setup function that will be run once before the actual test (scenarios) start and that can pass data to the test function(s). Since your first scenario runs only a single VU for a single iteration, you should move its implementation to the setup function instead.

More info and examples in the k6 test lifecycle docs

export let options = {
  scenarios: {
    employeeScenario: {
      executor: 'per-vu-iterations',
      exec: 'runEmployeeScenario',
      vus: 1,
      maxDuration: '10s',
    },
  },
};

export function setup() {
  // create resource(s) and collect ids(s) to array

  return ids;
}

export function runEmployeeScenario(ids) {
  // ids argument contains the result of the setup function.
}

Use the teardown function to clean up after your testrun has completed.

---

Alternatively, have a single scenario/function which performs both your calls:

export default function() {
  // create resource(s) and collect id(s) to array
  const ids = [ ... ];
  // use the ids from the array: for (id of ids) { ... }
}

This should work the same, given that both your scenarios run a single VU for exactly one iteration. Scenarios are most useful when you have independent VUs that perform independent operations in parallel with different request rates.