Kubetest

kubetest is a tool for running tests against a Kubernetes YAML or JSON configuration file. These tests can be used to enforce local or global best-practices, for example:

• Ensuring certain labels are set
• Prevent usage of images with the latest tag
• Prohibit privileged containers
• Enforce a naming convention for different resources

kubetest is currently alpha quality and undoutedly has a few issues. Things will change, hopefully for the better. Please open issues if you have feedback when trying it out.

Writing tests

Tests are written in Skylark, which is a small dialect of Python suitable for embedding in other programmes. This means you do not need an additional interpreter installed to run tests with kubetest. kubetest prioritises interopability over flexibility in this regard. Tests for Kubetest just require the kubetest binary to run. Let's take a look at an example test:

#// vim: set ft=python:
def test_for_team_label():
    if spec["kind"] == "Deployment":
        labels = spec["spec"]["template"]["metadata"]["labels"]
        assert_contains(labels, "team", "should indicate which team owns the deployment")

test_for_team_label()

Save the test file in a directory called tests, with an extension of .sky. You can change the default directory name using the --tests flag. You can now run kubetest against your configuration files.

$ kubetest my-deployment.yaml
WARN my-deployment.yaml Deployment should have at least 4 replicas
$ echo $?
1

If any of the tests fail then kubetest will return a non-zero exit code.

By default kubetest outputs information about failing tests only, but you can pass --verbose to get information about passing tests as well.

$ kubetest rc.yaml --verbose 
INFO rc.yaml should not use latest images
WARN rc.yaml ReplicationController should have at least 4 replicas

The spec variable

spec is a global variable passed into the Skylark code which contains the structure of the Kubernetes configuration passed in to kubetest. You'll need to be reasonably familiar with the structure of the Kubernetes API objects to write tests, but it is possible to write helper methods for common assertions.

Assertions

kubetest automatically makes available a set of assertions to make writing tests in Skylark more pleasant. A failed assertion results in kubetest exiting with a non-zero exit code, and assertions output results as shown above.

• assert_equal
• assert_contains
• assert_not_contains
• assert_not_equal
• assert_nil
• assert_not_nil
• fail
• fail_now
• assert_empty
• assert_not_empty
• assert_true
• assert_false

Assertions take zero, one or two arguments (noted above) depending on what they are comparing. They then take an additional message argument which is output when the assertion runs. For example the following assertion checks whether the variable labels contains the value team.

assert_contains(labels, "team", "should indicate which team owns the deployment")

Installation

Tagged versions of kubetest are built by Travis and automatically uploaded to GitHub. This means you should find tar.gz files under the release tab. These should contain a single kubetest binary for platform in the filename (ie. windows, linux, darwin). Either execute that binary directly or place it on your path.

wget https://github.com/garethr/kubetest/releases/download/0.1.0/kubetest-darwin-amd64.tar.gz
tar xf kubetest-darwin-amd64.tar.gz
cp kubetest /usr/local/bin

Windows users can download tar or zip files from the releases tab.

CLI

$ kubetest --help
Run tests against a Kubernetes YAML file

Usage:
  kubetest <file> [file...] [flags]

Flags:
  -h, --help           help for kubetest
      --json           Output results as JSON
  -t, --tests string   Test directory (default "tests")
      --verbose        Output passes as well as failures

Thanks

A big thank you goes to the authors of stretchr/testify from where much of the assertion code has been ported.