add an example to the kubectl plugin

[tpo]

SUMMARY

When looking at the kubectl plugin documentation I have no idea how to use the plugin. Having and example helps a lot. Please include an example such as the provided one. Thanks a lot for writing the plugin in the first place!

ISSUE TYPE

• Docs Pull Request

COMPONENT NAME

kubectl

[gravesm]

recheck

[gravesm]

Thanks for the PR @tpo. Instead of using kubectl here, can you instead use the kubernetes.core.k8s_info module?

[tpo]

@gravesm wrote:

Instead of using kubectl here, can you instead use the kubernetes.core.k8s_info module?

I'd like to. I just spent 90 minutes trying to get a simple k8s_info task to run to no avail. Same as with the kubectl plugin: the examples in the k8s_info docu are too dire to allow me to get it to work.

• Either k8s_info tries to first connect to the master node via ssh and access the k8s API there, which fails, because the credentials to access the API are not available on the node because my management machine is my own PC and not the master node.
• Or I am trying to get accessing the API from my local PC to work:

- hosts: master-node
  tasks:
    - local_action:
        module: k8s_info
        kind: Pod
        context: my_cluster
        # the following line doesn't seem to have any impact -
        # ansible will still try to connect to the k8s API on localhost
        # kubeconfig: /home/myself/.kube/config
      become: no
      register: pod_list
    - debug: var=pod_list

but am failing because for whatever reason, k8s_info is not accessing my default ~/.kube/config but instead trying to connect to the k8s API at localhost, which obviously fails, because localhost is not in fact a k8s cluster. Nota bene I'm running ansible 2.10 on Debian bullseye here.

So for the time being I'm stopping trying to get k8s_info to work and instead returning to my usual paid job duties again.

That means there are the following ways forward for me with this ticket:

• a) I get help on how to get k8s_info to work
• b) at some undefined time in the future I again take some time off from my normal schedule and try to resolve this twisted k8s_info mystery
• c) we leave the example as is
• d) I drop the retrieval of pod names from the example (which will obviously make it much less useful - because pod names are usually ephemeral) and put a static pod name into the example instead
• e) we scratch this pull request
• f) some other idea

What do you think? Do you have help or a suggestion on how to proceed?

[tpo]

Unfortunately it seems that I am unable to reserve myself time for option b) above during the coming 3 weeks or so. Any opinions on unstalling this pull request with either choice of a) c) d) e) or f) from above?

[gravesm]

recheck

[Akasurde]

recheck

[Akasurde]

recheck

[yurnov]

I have experience with kubernetes.core.kubectl connection plugin in production, I can create a better example if it help. I agree that looking into current documentation (i.e. for 3.0.0 or 3.1.0 tags) it's not clear how to use connection plugin at all.

[tpo]

I have experience with kubernetes.core.kubectl connection plugin in production, I can create a better example if it help.

Thank you, please do

[yurnov]

Hi @tpo and @gravesm, would you please review #741 that is my attempt to improve collection documentation by adding examples of using the collection plugin. Comment of @gravesm to use k8s_info considered

[tpo]

Hi @tpo and @gravesm, would you please review #741 that is my attempt to improve collection documentation by adding examples of using the collection plugin. Comment of @gravesm to use k8s_info considered

Thanks a lot for your very nice examples @yurnov - I have reviewed your pull request. If there's some formal syntax to tag your pull request as reviewed by myself then let me know and I can add that as well.

Thanks again, very appreciated!