kubectl multiple clusters how to best handle inventory.

[Matt Snoby]

Hi All,

I need some suggestions, I have an Ansible project that I use to deploy Kubernetes with KOPS.  The inventory file is very simple 

[targets]
localhost              ansible_connection=local

I also use this project to add additional functionality to my clusters with for instance the dashboard docker file.  Basically it looks like:

- name: Add Kubernetes dashboard deployment configuration
  template:
    src: kubernetes-dashboard.yaml.j2
    dest: /tmp/kubernetes-dashboard.yaml
    mode: 0644
  become: false


- name: Launch Kubernetes dashboard service
  command: "kubectl create -f /tmp/kubernetes-dashboard.yaml"

However I was thinking it would be better if I could pass a context into the kubectl command like cluster foo.east.example.com or foo.west.example.com.  This is all supported already in kubectl like this:

- name: Launch Kubernetes dashboard service
  command: "kubectl --context=foo.east.example.com create -f /tmp/kubernetes-dashboard.yaml"

What I'm not sure about is how I can setup my inventory file to perhaps have variables such as the cluster name.  For instance could I do something like

[foo.east.example.com]
localhost              ansible_connection=local



[foo.west.example.com]
localhost              ansible_connection=local

And then setup a group/foo.east.example.com.yaml file and put my variables in there for that cluster and do the same for a group/foo.west.example.com/yaml file ?

Is that a best practices way to do it?

Thanks for any insight you can provide!

Matt