Testing

Because Pulumi uses general purpose programming languages to provision cloud resources, you can take advantage of native testing frameworks and perform automated tests of your infrastructure.

There are multiple styles of automated testing: Unit testing of individual pieces, integration testing of system components, end-to-end testing of the complete application—each has its own advantages and appropriate uses.

Similarly, Pulumi provides multiple testing styles for cloud programs:

  • Unit Tests are fast in-memory tests that mock all external calls.
  • Property Tests run resource-level assertions while infrastructure is being deployed.
  • Integration Tests deploy ephemeral infrastructure and run external tests against it.

The following table summarizes the differences between the three approaches:

Unit TestsProperty TestsIntegration Tests
Provision real infrastructureNoYesYes
Require the Pulumi CLINoYesYes
Time to executeMillisecondsSecondsMinutes
LanguageSame as Pulumi programNode.js or PythonAny language
Validation targetResource inputsResource inputs and outputsExternal endpoints

We encourage you to try all three styles of testing and use the ones that are most suitable for your quality targets, development practices, and application style. For many teams, a combination of these approaches may make sense: unit tests that can run quickly and validate program logic, property tests to validate key correctness invariants, and integration tests to test end-to-end interaction of infrastructure components.

Unit Testing

Unit tests evaluate the behavior of your code in isolation, while all external dependencies are replaced by mocks. Unit tests run in memory without any out-of-process calls, which makes them blazingly fast. Therefore, unit tests are suitable for fast feedback loops during development, including Test-Driven Development (TDD).

Unit tests are authored in the same language as the Pulumi program under test. You can use your favorite test and mock frameworks like Mocha for Node.js or NUnit for .NET.

Because cloud resources are not created, you can’t write a test that would evaluate the behavior of infrastructure. For example, you can’t make HTTP requests to endpoints, because there’s no webserver to serve them.

Learn more and get started with Unit Testing.

Property Testing

Property tests are based on Policy as Code (also known as “CrossGuard”), Pulumi’s offering to set guardrails and enforce compliance for cloud resources. In addition to authoring company-wide policies, CrossGuard enables another type of infrastructure testing. Each policy becomes a property, an invariant, that a test evaluates and asserts.

Property tests run inside the Pulumi CLI before and after infrastructure provisioning. In contrast to “black-box” integration testing, policy rules have access to all input and output values of all cloud resources in the stack. As opposed to unit testing, property tests can evaluate real values returned from the cloud provider instead of the mocked ones.

Property tests can run against any cloud environment: it can be a persistent “acceptance” stack, an ephemeral cloud environment created for each pull request, or a combination of those.

Learn more and get started with Property Testing.

Integration Testing

Integration testing takes a different approach of unit tests: the tests deploy cloud resources and validate their actual behavior.

An integration test invokes the Pulumi command-line interface (CLI) to deploy infrastructure to an ephemeral environment. When the resources are deployed, the test retrieves endpoints of the infrastructure from the stack outputs: usually, a URL or a public IP address. The test verifies that the infrastructure behaves as expected; for example, it expects a valid HTML document from a health-check endpoint or runs a suite of application-level tests against the public API. When the tests finish, the infrastructure is destroyed.

The great advantage of integration tests is the ability to test the actual cloud infrastructure and its real properties. However, compared to unit tests, integrations tests take more time to execute.

Depending on the type of resources and frequency of testing, even short-lived ephemeral environments may incur notable charges from the cloud provider. Be sure to plan accordingly and measure frequently.

Learn more and get started with Integration Testing.

Examples

Several complete and runnable tests are available in the examples repository:

ExampleDescription
Unit Tests in TypeScriptMock-based unit tests in TypeScript.
Unit Tests in PythonMock-based unit tests in Python.
Unit Tests in GoMock-based unit tests in Go.
Unit Tests in C#Mock-based unit tests in C#.
Property Testing in TypeScriptTests based on Policy-as-Code in TypeScript.
Integration Testing in GoDeploy-check-destroy tests in Go.