* Initialize OpenshiftGroupsSync attributes early
The fail_json() method calls close_connection(), but this can fail when
the python-ldap library is not installed as close_connection() gets
called before the __ldap_connection attribute has been defined.
* Use already defined virtualenv_command
* Fix tests
* Add changelog fragment
* Fix integration tests for k8s 2.3
Previous to
https://github.com/ansible-collections/kubernetes.core/pull/408 an
attempt to list resources for which the user was not authorized resulted
in an unhandled exception. The task now does not fail, but instead
reports the error message.
* try to fix sanity
* Fix tests
* fix tests
* fix tests
* Revert change to downstream build
* fix ocp auth adding / to oauth discovery url
* fix space after comma
* add changelog fragment
* Convert line endings
Co-authored-by: Mike Graves <mgraves@redhat.com>
* relax the dependency with kubernetes.core
We can use the collection with the 2.2 branch of kubernetes.core.
* galaxy: use <2.3.0 has the upper version bound
See: https://github.com/ansible/ansible-zuul-jobs/pull/1119
- delay the loading of external modules when possible
- delay the loading of OKDRawModule and OpenShiftProcess classes
- k8s reuse the design of the kubernetes.core modules
We've got a chicken/egg problem that prevent us from properly
reporting if kubernetes.core is missing. We need args_common to create
the module object. And we need the module object to report the missing
dependency. The dependency is declared in the galaxy.yml file anyway,
the problem should not happen.
Follow up to the recent upgrade of pylint in the devel branch, we've
got a new use-a-generator error in k8s.py:
Use a generator instead 'all(desired.get(key, True) == item.get(key, False) for key in keys)'
This commit disables the error until we've got time to rewrite this
part.
Also, since the older branches still use an ancient version of pylint,
we've got another message saying the error does not exist (yet!).
This commit also adds the right entries in the ignore files.
Address the following warning:
```
ERROR: Found 1 shellcheck issue(s) which need to be resolved:
ERROR: ci/incluster_integration.sh:16:7: SC2236: Use -n instead of ! -z.
```
* Enable turbo mode
This enables turbo mode (default is off) for the modules in this
collection that use the client from the kubernetes.core collection.
* Add changelog fragment
* Fix broken links for downstream collection
The links to the repo on Automation Hub are broken for the downstream
collection. This skips the replacement step for galaxy.yml.
* Add changelog fragment
* Update to work with k8s 2.0
This makes the necessary changes to get the collection working with
kubernetes.core 2.0. The biggest changes here will be switching from the
openshift client to the kubernetes client, and dropping Python 2
support.
* Install kubernetes not openshift
* Add changelog fragment
* Clean up CI build
There were a few problems with the existing build process, such as, the
python version not being passed during sanity tests and versions being
hard coded elsewhere. This change ensures that each test scenario gets
the correct parameters passed through. It also generally cleans up the
process for building the test environment to let ansible-galaxy do its
thing. This will make sure the correct version of dependencies get
installed.
* Add changelog fragment
