Scorecard

Overview

The scorecard command, part of the operator-sdk, executes tests on your operator based upon a configuration file and test images.

Tests are implemented within test images that are configured and constructed to be executed by scorecard.

Scorecard assumes it is being executed with access to a configured Kubernetes cluster. Each test is executed within a Pod by scorecard, from which pod logs are aggregated and test results sent to the console.

Scorecard has built-in basic and OLM tests, and it also provides a means to execute custom test definitions.

Requirements

The scorecard tests make no assumptions as to the state of the operator being tested. Creating operators and custom resources for an operator are left outside the scope of the scorecard itself.

Scorecard tests can however create whatever resources they require if the tests are designed for resource creation.

Running the Scorecard

  1. A default set of kustomize files should have been scaffolded by operator-sdk init. If that is not the case, run operator-sdk init as you would have to initialize your project and copy scaffolded files:
  1. $ TMP_PROJECT="$(mktemp -d)/<current-project-name>"
  2. $ mkdir "$TMP_PROJECT"
  3. $ pushd "$TMP_PROJECT"
  4. $ operator-sdk init
  5. $ popd
  6. $ cp -r "$TMP_PROJECT"/config/scorecard ./config/

The default config generated by this kustomization can be immediately run against your operator. See the config file section for an explanation of the configuration file format.

  1. (Re)enerate bundle manifests and metadata for your Operator. make bundle will automatically add scorecard annotations to your bundle’s metadata, which is used by the scorecard command to run tests.
  2. Execute the scorecard command. See the command args section for an overview of command invocation.

Configuration

The scorecard test execution is driven by a configuration file named config.yaml, generated by make bundle. The configuration file is located at the following location within your bundle directory (bundle/ by default):

  1. $ tree ./bundle
  2. ./bundle
  3. ...
  4. └── tests
  5. └── scorecard
  6. └── config.yaml

Config File

The following YAML spec is an example of the scorecard configuration file:

  1. kind: Configuration
  2. apiversion: scorecard.operatorframework.io/v1alpha3
  3. metadata:
  4. name: config
  5. stages:
  6. - parallel: true
  7. tests:
  8. - image: quay.io/operator-framework/scorecard-test:latest
  9. entrypoint:
  10. - scorecard-test
  11. - basic-check-spec
  12. labels:
  13. suite: basic
  14. test: basic-check-spec-test
  15. - image: quay.io/operator-framework/scorecard-test:latest
  16. entrypoint:
  17. - scorecard-test
  18. - olm-bundle-validation
  19. labels:
  20. suite: olm
  21. test: olm-bundle-validation-test

The configuration file defines the tests that scorecard executes. Tests are grouped into stages for fine-grained control of parallelism. The following fields of the scorecard configuration file define the test as follows:

Config FieldDescription
imagethe test container image name that implements a test
entrypointthe command and arguments that are invoked in the test image to execute a test
labelsscorecard-defined or custom labels that select which tests to run

Command Args

The scorecard command has the following syntax:

  1. $ operator-sdk scorecard <bundle_dir_or_image> [flags]

The scorecard requires a positional argument that holds either the on-disk path to your operator bundle or the name of a bundle image.

For further information about the flags see the CLI documentation.

Parallelism

The configuration file allows operator developers to define separate stages for their tests. Stages run sequentially in the order they are defined in the configuration file. A stage contains a list of tests and a configurable parallel setting.

By default (or when a stage explicitly sets parallel to false), tests in a stage are run sequentially in the order they are defined in the configuration file. Running tests one at a time is helpful to guarantee that no two tests interact and conflict with each other.

However, if tests are designed to be fully isolated, they can be parallelized. To run a set of isolated tests in parallel, include them in the same stage and set parallel to true. All tests in a parallel stage are executed simultaneously, and scorecard waits for all of them to finish before proceding to the next stage. This can make your tests run much faster.

Selecting Tests

Tests are selected by setting the --selector CLI flag to a set of label strings. If a selector flag is not supplied, then all the tests within the scorecard configuration file are executed.

Tests are executed serially, one after the other, with test results being aggregated by scorecard and written to stdout.

To select a single test (basic-check-spec-test) you would enter the following:

  1. $ operator-sdk scorecard <bundle_dir_or_image> -o text --selector=test=basic-check-spec-test

To select a suite of tests, olm in this case, you would specify a label that is used by all the OLM tests:

  1. $ operator-sdk scorecard <bundle_dir_or_image> -o text --selector=suite=olm

To select multiple tests, you could specify them as follows:

  1. $ operator-sdk scorecard <bundle_dir_or_image> -o text --selector='test in (basic-check-spec-test,olm-bundle-validation-test)'

Built-in Tests

The scorecard ships with pre-defined tests that are arranged into suites.

Basic Test Suite

TestDescriptionTest Name
Spec Block ExistsThis test checks the Custom Resource (CRs) created in the cluster to make sure that all CRs have a spec block.basic-check-spec-test

OLM Test Suite

TestDescriptionShort Name
Bundle ValidationThis test validates the bundle manifests found in the bundle that is passed into scorecard. If the bundle contents contain errors, then the test result output will include the validator log as well as error messages from the validation library. See this document for details on bundles.olm-bundle-validation-test
Provided APIs have validationThis test verifies that the CRDs for the provided CRs contain a validation section and that there is validation for each spec and status field detected in the CR.olm-crds-have-validation-test
Owned CRDs Have Resources ListedThis test makes sure that the CRDs for each CR provided via the cr-manifest option have a resources subsection in the [owned CRDs section][owned-crds] of the CSV. If the test detects used resources that are not listed in the resources section, it will list them in the suggestions at the end of the test. Users are required to fill out the resources section after initial code generation for this test to pass.olm-crds-have-resources-test
Spec Fields With DescriptorsThis test verifies that every field in the Custom Resources’ spec sections have a corresponding descriptor listed in the CSV.olm-spec-descriptors-test
Status Fields With DescriptorsThis test verifies that every field in the Custom Resources’ status sections have a corresponding descriptor listed in the CSV.olm-status-descriptors-test

Scorecard Output

The --output flag specifies the scorecard results output format.

JSON format

See an example of the JSON format produced by a scorecard test:

  1. {
  2. "apiVersion": "scorecard.operatorframework.io/v1alpha3",
  3. "kind": "TestList",
  4. "items": [
  5. {
  6. "kind": "Test",
  7. "apiVersion": "scorecard.operatorframework.io/v1alpha3",
  8. "spec": {
  9. "image": "quay.io/operator-framework/scorecard-test:latest",
  10. "entrypoint": [
  11. "scorecard-test",
  12. "olm-bundle-validation"
  13. ],
  14. "labels": {
  15. "suite": "olm",
  16. "test": "olm-bundle-validation-test"
  17. }
  18. },
  19. "status": {
  20. "results": [
  21. {
  22. "name": "olm-bundle-validation",
  23. "log": "time=\"2020-06-10T19:02:49Z\" level=debug msg=\"Found manifests directory\" name=bundle-test\ntime=\"2020-06-10T19:02:49Z\" level=debug msg=\"Found metadata directory\" name=bundle-test\ntime=\"2020-06-10T19:02:49Z\" level=debug msg=\"Getting mediaType info from manifests directory\" name=bundle-test\ntime=\"2020-06-10T19:02:49Z\" level=info msg=\"Found annotations file\" name=bundle-test\ntime=\"2020-06-10T19:02:49Z\" level=info msg=\"Could not find optional dependencies file\" name=bundle-test\n",
  24. "state": "pass"
  25. }
  26. ]
  27. }
  28. }
  29. ]
  30. }

Text format

See an example of the text format produced by a scorecard test:

  1. --------------------------------------------------------------------------------
  2. Image: quay.io/operator-framework/scorecard-test:latest
  3. Entrypoint: [scorecard-test olm-bundle-validation]
  4. Labels:
  5. "suite":"olm"
  6. "test":"olm-bundle-validation-test"
  7. Results:
  8. Name: olm-bundle-validation
  9. State: pass
  10. Log:
  11. time="2020-07-15T03:19:02Z" level=debug msg="Found manifests directory" name=bundle-test
  12. time="2020-07-15T03:19:02Z" level=debug msg="Found metadata directory" name=bundle-test
  13. time="2020-07-15T03:19:02Z" level=debug msg="Getting mediaType info from manifests directory" name=bundle-test
  14. time="2020-07-15T03:19:02Z" level=info msg="Found annotations file" name=bundle-test
  15. time="2020-07-15T03:19:02Z" level=info msg="Could not find optional dependencies file" name=bundle-test

NOTE The output format spec for each test matches the Test type layout.

Exit Status

The scorecard return code is 1 if any of the tests executed did not pass and 0 if all selected tests pass.

Extending the Scorecard with Custom Tests

Scorecard will execute custom tests if they follow these mandated conventions:

  • tests are implemented within a container image
  • tests accept an entrypoint which include a command and arguments
  • tests produce v1alpha3 scorecard output in JSON format with no extraneous logging in the test output
  • tests can obtain the bundle contents at a shared mount point of /bundle
  • tests can access the Kubernetes API using an in-cluster client connection

See here for an example of a custom test image written in Go.

Writing custom tests in other programming languages is possible if the test image follows the above guidelines.

Last modified September 16, 2020: [1.0.x] Fix broken links (#3893) (cfabe8d6)