geerlingguy.k8s_manifests

Ansible Role: K8s Manifests

This Ansible Role allows you to apply Kubernetes manifests (which can be either templates or direct files) to Kubernetes clusters.

Requirements

You need the Pip package: openshift
If you're running this on your own machine (e.g., with connection: local), you might need to set ansible_python_interpreter: "{{ ansible_playbook_python }}" for it to work correctly.

Role Variables

Below are the available variables along with their default values (check defaults/main.yml):

k8s_manifests:
  # Add the path to manifests inside `k8s_manifests_base_dir`.
  - monitoring/prometheus

  # Use `file` lookup if you prefer not to template the manifest.
  - dir: monitoring/grafana-configmap
    lookup_type: 'file'

  # Set a namespace for each manifest if needed.
  - dir: docker-registry
    namespace: registry

This is a list of Kubernetes manifest directories to apply to your Kubernetes cluster. You can specify either simple paths or a dictionary with dir (directory or folder name), optional lookup_type (type of lookup for the manifest.yml file), and optional namespace (set to manifest_namespace).

The role will look inside each directory for a manifest and apply the manifest.yml file (and its contents) using the Ansible k8s module.

By default, the role will treat the manifest.yml file as a template (and it will automatically include any variables from a vars.yml file next to it). However, you can skip templating by setting lookup_type: file.

k8s_manifests_base_dir: '' # should end with /, e.g., 'base_dir/'

If you set this string, it will be added to each dir/path listed in k8s_manifests. This is helpful if your manifests are stored outside the Ansible playbook directory.

k8s_manifests_state: present

This determines if the k8s module should have state as present (apply the manifest) or absent (remove it). Note that absent doesn't always delete all Kubernetes resources in a manifest.

k8s_force: false

If set to true and k8s_manifests_state is present, it will replace existing objects. By default, Ansible and Kubernetes will update (patch) the resource.

k8s_kubeconfig: ~/.kube/config

This specifies the path to the kubeconfig file for connecting to the Kubernetes cluster.

k8s_resource_namespace: ''
k8s_manage_namespace: true

By default, the role assumes you'll deploy resources into a specific namespace. If you set k8s_resource_namespace, the role will ensure that namespace exists before applying any manifests. You can turn off namespace management by setting k8s_manage_namespace: false.

k8s_no_log: true

This controls whether to log the details of each manifest application to Ansible output. Since sensitive data could be included, it is set to be secure by default. Change to false for debugging.

Dependencies

None.

Example Playbooks

Simple Example - Running on Localhost

---
- hosts: localhost
  connection: local
  gather_facts: no

  vars:
    ansible_python_interpreter: "{{ ansible_playbook_python }}"
    k8s_kubeconfig: ~/.kube/config-my-cluster
    k8s_manifests_base_dir: k8s-manifests/
    k8s_manifests:
      - storageclass

  roles:
    - role: geerlingguy.k8s_manifests

Check the k8s-manifests directory and its README for an example of a templated manifest layout with a vars file.

Running as Part of a Larger Play

---
- hosts: k8s_cluster
  become: true

  vars:
    ansible_python_interpreter: python
    k8s_manage_namespace: false
    k8s_no_log: false
    k8s_manifests_base_dir: k8s-manifests/
    k8s_manifests:
      - storageclass
      - dir: docker-registry
        namespace: registry

  tasks:
    - name: Set the python interpreter correctly.
      set_fact:
        ansible_python_interpreter: "{{ ansible_playbook_python }}"

    - import_role:
        name: geerlingguy.k8s_manifests
      tags: ['kubernetes', 'nfs', 'drupal', 'registry']
      delegate_to: localhost
      become: false
      run_once: true