Closed a-meynard closed 3 years ago
@a-meynard We appreciate the feedback. It is not ideal so I wouldn't call it expected. It's more of an overlooked use case in our initial work.
This sounds like issue #307 that was fixed in PR #320. While this functionality has been merged, we haven't made a release with it. That will be coming once we get into the new year.
Is this the same issue as #307 that we can close this or is it something more?
Hello @tima
I think my issue is different from #307, but it seems that #320 will solve it I'll give it a try tomorrow and close this issue if it's solved by the PR :)
the bug does not appear when using the main branch. I assume #320 resolves the problem Thank you @tima I will close this Issue and wait for a new release of this repo (I'm using 1.1.1 right now)
SUMMARY
I passed my playbooks from community.kubernetes == 0.11.1 to 1.1.1 and the
kubeconfig
parameter stopped working correctly. This parameter used to search for kubeconfig on the Ansible Target but it seems that it is now searching for kubeconfig path on the Ansible Controller.ISSUE TYPE
COMPONENT NAME
community.kubernetes.k8s
ANSIBLE VERSION
CONFIGURATION
OS / ENVIRONMENT
I use kubernetes==11.0.0 because of issue #314
STEPS TO REPRODUCE
Using this playbook, having the kubeconfig file on remote at path
/root/.kube/config
with a context calledkubernetes-admin@kubernetes
.EXPECTED RESULTS
Create the test namespace by using kubeconfig file from the remote.
ACTUAL RESULTS
The module search for kubeconfig file on the Ansible Controller host (which has no kubeconfig files) so it fails by saying that ansible controller has nos valid kubeconfig file at path
/root/.kube/config
WORKAROUND
Actually, using this playbook (with environment variable) works fine, but I don't know if this is expected. If this is expected, I think that the documentation should mention that the use of environmental variables implies a different behaviour.