Changelog

Based on Keep a Changelog, following Semantic Versioning.

Unreleased

1.3.2 - 2024-10-31

fixed
- _deployment_running which currently fails for deployments without unavailableReplicas set if they run without issues.

1.3.1 - 2024-10-23

changed
- remove support for Python 3.8 and 3.9
- add support for Python 3.12 and 3.13
fixed
- the _deployment_running method was missing a check for unavailableReplicas

1.2.0 - 2024-02-06

added:
- make flux resource ready state check timeouts configurable, still default to 30s

1.1.2 - 2023.09.13

changed:
- dependency updates

[1.1.1] - 2023.03.08

fixed:
- the kubectl() method was not capturing standard error stream
changed:
- the kubectl() method allows now to be executed within a shell (parameter use_shell, False by default)

[1.0.2] - 2022.10.19

fixed:
- remove defaults for cmd line options - they were breaking env variable options loading

[1.0.1] - 2022.08.03

changed:
- _flux_cr_ready made public by renaming to flux_cr_ready
- Test information data is now primarily being input by (partially mandatory) environment variables. Majority of old command line parameters should still work, but environment variables usage is encouraged. This allows pytest-helm-charts to be easily used without the app-test-suite project, which was enforcing the cmd line parameters before. The following env vars are recognised and used in fixtures (when requested):
  - "KUBECONFIG" - (mandatory) a path to kube config file used to connect to a k8s cluster
  - "ATS_CHART_PATH" - path to a chart being tested (if a chart is tested)
  - "ATS_CHART_VERSION" - version of the chart being tested (if a chart is tested)
  - "ATS_CLUSTER_TYPE" - (informative only) type of the cluster used for testing
  - "ATS_CLUSTER_VERSION" - (informative only) k8s version of the cluster used for testing
  - "ATS_APP_CONFIG_FILE_PATH" - optional path to a values.yaml file used to configure a chart under test (if a chart is tested)
  - "ATS_EXTRA_*" - any such arbitrary variable value will be extracted and included in the test_extra_info fixture
- chart_extra_info fixture was removed, as the more general test_extra_info is available
updated:
- pytest upgraded from 6.x series to 7.x

[0.7.1]

fixed:
- 'repositories' attribute of Catalog CR is now handled

[0.7.0]

changed:
- all the changes from beta releases below
- fixed docs issues

[0.7.0-beta.2]

changed:
- rename 'api' module to 'k8s'
- update Kustomization to v1beta2
fixed:
- in HelmRelease, ensure 'targetNamespace' if it doesn't exist
- logging level adjusted (debug only if no errors) and added consistent log messages about creating and deleting resources
- add missing imports in plugin.py

[0.7.0-beta.1]

added:
- all kubernetes API Objects targeting fixtures have now a function scoped counterpart named *_function_scope
- package flux was added to deal with Flux CRs
changed:
- all fixture factories that create objects and delete them after they are out of scope, now actively wait for the object to be gone. Previously, only delete() request was sent without checking if the object is gone, which was causing a bunch of race conditions between test runs.
- All recursively-required missing resources are now automatically created. This usually means that if the Namespace you want to put your resource in doesn't exist, it will be created automatically for you. The same applies to App and the required Catalog.
- switch from Callable to Protocol for factory types (much better type hinting)
- all functions making API objects take now optional extra_metadata and extra_spec arguments, which are merged with object definitions without any restrictions nor validation
- wait_for_apps_to_run accepts now a new parameter fail_fast: bool = False; when it's True and the App's status ever reaches failed, the wait fails as well, without waiting for subsequent state changes.
- cleanups:
  - all methods creating objects, but not submitting them to k8s API, match now make-*-object pattern
  - HTTPClient connection object is always the first method parameter
  - removed the namespaceConfig* arguments of make_app_object, as they can be now included using the extra-* args
  - multiple classes were moved across modules and packages to match the following rules:
    - every Resource / CustomResource is a single module
    - groups of API Resources (like giant swarm app platform or flux - in the future) go to packages
  - wait_for_namespaced_objects_condition function is renamed to wait_for_objects_condition as it now supports both cluster-scope and namespace-scope resources. Additionally, it allows now to pass a function for checking for fail fast conditions in objects awaited.
fixed:
- namespaces requested from namespace_factory* fixtures are now deleted only if they were created by the fixture (and not already existing in the cluster)

[0.6.0]

breaking change: all App CR related functions and fixtures that require information about the Catalog the App references need now also a new catalog_namespace parameter

[0.5.3]

add support for Catalog CR from Giant Swarm App Platform
mark AppCatalog CR related functions and classes as deprecated

[0.5.2]

fix docstring formatting
make get_app_catalog_object function public

[0.5.1]

added: a couple of extracted helper methods to use outside of pytest's fixture mechanism:
- create_app
- delete_app
- wait_for_app_to_be_deleted
unified imports to absolute paths

[0.5.0]

fix: wait_for_deployments_to_run needs to check for availablePods as well
added: namespace_factory fixture you can use to create test time only namespaces
change: app_factory takes now one more kwarg deployment_namespace, which configures the target deployment namespace for the created App CR; in other words, the namespace where the App CR is created can be now different than the deployment_namespace where app's components are deployed
change: if the namespace passed as namespace argument to app_factory doesn't exist, it is automatically created using the new namespace_factory fixture

[0.4.1]

fix: work-around a bug in a depended lib that was making wait_for_deployments_to_run fail

[0.4.0]

change: app_factory_func now checks if the passed App deploys successfully. This is configured using the timeout_sec argument, which by default is equal 60. Use value 0 to disable checking if App deploys OK.
added:
- wait_for_apps_to_run helper method
- wait_for_daemon_sets_to_run helper method
- kubectl method in kube_cluster, which allows you to make any calls using the binary kubectl
- random_namespace fixture that creates a temporary namespace with a random name

[0.3.2]

added: support for namespacedConfig for App CR (app_factory_func fixture)

[0.3.1]

Remove any reference related to supporting multiple clusters. From now on this project always expects to get a kube.config for an already existing cluster.

[0.2.1]

added: include python 3.9 as a version for test running
changed: updated dependencies
removed: redundant code that was accepted in pykube-ng lib

0.1.10

added: more functions to help with handling namespaces, jobs and stateful sets.

0.1.9

fix: Don't supply port from kwargs to requests in util.proxy_http_request

0.1.8

change: use loadtest app helm chart 0.2.0

0.1.7

fix: type definitions for wait_for_namespaced_object_condition function

0.1.6

fix: fix the chart_extra_info fixture and add an example showing how to use it.

0.1.5

add: --chart-extra-info CLI option that can be used to pass arbitrary key=value maps to the test cases. Values fo the map are accessible in code as chart_extra_info fixture.

0.1.4

fix: wait_for_namespaced_objects_conditions was exiting prematurely when the objects were not yet present

0.1.3

fix plugin layout - import of fixtures
fix the gatling fixture
add more helper methods
- wait for job / deployment
- api proxy access

0.1.2 2020-07-07

no code changes
tools include pre-commit

0.1.1 2020-07-06

First release version

0.1.0 2020-05-03

Initial commit