Ansible Operator

Ansible Operator

Create a new directory for your Ansible-Operator project.

mkdir ~/ansible-operator
cd ~/ansible-operator

Use the CLI to create a new Ansible-based Operator project with the new command. operator-sdk new --type ansible has two required flags --api-version and --kind. These flags are used to generate proper Custom Resource files and an Ansible Role whose name matches the input for --kind.

operator-sdk new memcached-operator --kind=Memcached --type=ansible --skip-git-init
cd memcached-operator

To speed things up, we can reuse a role that has been written inside of our operator. The role we will use is dymurray.memcached_operator_role. The section at the bottom of this document will also show the reader how to create that Ansible Role from scratch. To get started, install the Ansible Role inside of the project.

ansible-galaxy install dymurray.memcached_operator_role -p ./roles

Delete generated scaffolding role

rm -rf ./roles/Memcached

This role provides the user with a variable size which is an integer to control the number of replicas to create. You can find the default for this variable in roles/dymurray.memcached_operator_role/defaults/main.yml.

cat roles/dymurray.memcached_operator_role/defaults/main.yml

View the Ansible role.

cat roles/dymurray.memcached_operator_role/tasks/main.yml

Update the watches.yaml file

cat > watches.yaml <<EOF
- version: v1alpha1
  kind: Memcached
  role: /home/vagrant/ansible-operator/memcached-operator/roles/dymurray.memcached_operator_role

Before running the operator, Kubernetes needs to know about the new custom resource definition the operator will be watching.

kubectl create -f deploy/crds/cache_v1alpha1_memcached_crd.yaml

Run the operator locally with the default kubernetes config file present at $HOME/.kube/config.

operator-sdk up local --namespace myproject

Create the Memcached CR.

kubectl create -f deploy/crds/cache_v1alpha1_memcached_cr.yaml