In the lifecycle of a k6 test, a script always runs through these stages in the same order:
- Code in the init context prepares the script, loading files, importing modules, and defining the test lifecycle functions. Required.
- The setup function runs, setting up the test environment and generating data. Optional.
- VU code runs in the default or scenario function, running for as long and as many times as the options define. Required.
- The teardown function runs, postprocessing data and closing the test environment. Optional.
Except for init code, each stage occurs in a lifecycle function, a function called in a specific sequence in the k6 runtime.
For examples and implementation details of each stage, refer to the subsequent sections.
|1. init||Load local files, import modules, declare lifecycle functions||Open JSON file, Import module||Once per VU*|
|2. Setup||Set up data for processing, share data among VUs||Call API to start test environment||Once|
|3. VU code||Run the test function, usually default||Make https requests, validate responses||Once per iteration, as many times as the test options require|
|4. Teardown||Process result of setup code, stop test environment||Validate that setup had a certain result, send webhook notifying that test has finished||Once **|
* In cloud scripts, init code might be called more often.
** If the Setup function ends abnormally (e.g throws an error), the teardown() function isn't called. Consider adding logic to the setup() function to handle errors and ensure proper cleanup.
The init stage is required. Before the test runs, k6 needs to initialize the test conditions. To prepare the test, code in the init context runs once per VU.
Some operations that might happen in init include the following:
- Import modules
- Load files from the local file system
- Configure the test for all options
- Define lifecycle functions for the VU, setup, and teardown stages (and for custom or handleSummary() functions, too).
All code that is outside of a lifecycle function is code in the init context. Code in the init context always executes first.
Separating the init stage from the VU stage removes irrelevant computation from VU code, which improves k6 performance and makes test results more reliable. One limitation of init code is that it cannot make HTTP requests. This limitation ensures that the init stage is reproducible across tests (the response from protocol requests is dynamic and unpredictable)
Scripts must contain, at least, a scenario function that defines the logic of the VUs. The code inside this function is VU code. Typically, VU code is inside the default function, but it can also be inside the function defined by a scenario (see subsequent section for an example).
VU code runs over and over through the test duration. VU code can make HTTP requests, emit metrics, and generally do everything you'd expect a load test to do. The only exceptions are the jobs that happen in the init context.
- VU code does not load files from your local filesystem.
- VU code does not import any other modules.
Again, instead of VU code, init code does these jobs.
A VU executes the default() function from start to end in sequence. Once the VU reaches the end of the function, it loops back to the start and executes the code all over.
As part of this "restart" process, k6 resets the VU. Cookies are cleared, and TCP connections might be torn down (depending on your test configuration options).
Like default, setup and teardown functions must be exported functions. But unlike the default function, k6 calls setup and teardown only once per test.
- setup is called at the beginning of the test, after the init stage but before the VU stage.
- teardown is called at the end of a test, after the VU stage (default function).
You can call the full k6 API in the setup and teardown stages, unlike the init stage. For example, you can make HTTP requests:
You can skip the execution of setup and teardown stages using the options --no-setup and --no-teardown.
Again, let's have a look at the basic structure of a k6 test:
You might have noticed the function signatures of the default() and teardown() functions take an argument, referred to here as data.
Here's an example of passing some data from the setup code to the VU and teardown stages:
For example, with the data returned by the setup() function, you can:
- Give each VU access to an identical copy of the data
- Postprocess the data in teardown code
However, there are some restrictions.
- You can pass only data (i.e. JSON) between setup and the other stages. You cannot pass functions.
- If the data returned by the setup() function is large, it will consume more memory.
- You cannot manipulate data in the default() function, then pass it to the teardown() function.
It's best to think that each stage and each VU has access to a fresh "copy" of whatever data the setup() function returns.
It would be extremely complicated and computationally intensive to pass mutable data between all VUs and then to teardown, especially in distributed setups. This would go against a core k6 goal: the same script should be executable in multiple modes.
k6 has a few additional ways to use lifecycle functions:
handleSummary(). If you want to make a custom summary, k6 calls one more lifecycle function at the very end of the test.
For details, refer to Custom summary.
Scenario functions. Instead of the default function, you can also run VU code in scenario functions.scenario-function.js