Introduction
In my last post, I went over linting Ansible playbooks manually. In this post, I’m going to give you a brief introduction on how to lint playbooks automatically, using Drone.
I was inspired to post this after watching Jeff Geerling’s (geerlingguy on Github) Ansible 101 YouTube series. In it, he mentions using automated testing/linting of his playbooks. If you haven’t seen it, give it a watch.
As with before, I’m not a developer or professional DevOps person. I know just enough to be dangerous, but this should be enough to get you going.
What is Drone?
Drone is continuous integration and continuous delivery/deployment (often written as CI/CD) software. It allows you to take automated actions when code is pushed to a repository (e.g., test the code, push to a Docker registry, push to production, alert if failed, etc…).
There are many popular CI/CD solutions, including, but not limited to:
However, most of these solutions were overkill for my workload. Considering I’m a noob at this, I was looking for something lightweight, minimal, and most importantly, easy to setup. Drone fit my requirements perfectly.
Why automate linting?
As I said in my last post, I lint my playbooks because I’d like to have some confidence that updates to my playbooks will work, without having to run them after every update. However, I currently have to manually lint any playbooks after updates are made.
Since my Ansible playbooks are all checked into git, I can setup this workflow.
- My self-hosted Gitea instance notifies Drone of changes via a webhook
- Drone spins up a Docker container of a pre-built Ubuntu image with Ansible installed
- Drone runs a command to lint my playbooks using that container’s Ansible installation
Installation
Version Control System
You need some sort of version control system (VCS) that supports webhooks. I’m using Gitea at home, running on Docker. Drone supports providers other than Gitea, including GitHub, GitLab, Gogs, and Bitbucket.
Note - Drone needs to be notified by your VCS via a webhook. For me, both Gitea and Drone are self-hosted at home. If you’re using GitHub (or other public VCS), you need to either:
- expose your self-hosted Drone instance to the world (e.g., put your Drone instance in your network’s DMZ)
- host Drone yourself in the cloud (e.g., on a VPS)
- use the hosted instance of Drone
OAuth key
Once you select a VCS provider, you’ll need to create an OAuth key. This will allow you to login to Drone using your VCS provider of choice.
The instructions will be different for each provider, so I’ll just link the top-level installation page here.
Drone
Everything in Drone is Docker-first. Therefore, Drone is designed to be installed in a Docker container. Below is a sanitized version of my docker-compose.yml file to run Drone and the Docker runner (the runner is the application that actually executes the CI/CD pipeline steps). By Default, Drone uses a SQLite database, but also supports PostgreSQL and MySQL as well. I didn’t post my PostgreSQL configuration here to keep things simple.
version: '3'
services:
drone:
container_name: drone
image: drone/drone:1
restart: unless-stopped
environment:
- DRONE_SERVER_HOST=drone.internal.mydomain.com
- DRONE_SERVER_PROTO=https
- DRONE_GITEA_SERVER=https://git.internal.mydomain.com
- DRONE_GITEA_CLIENT_ID=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
- DRONE_GITEA_CLIENT_SECRET=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
- DRONE_RPC_SECRET=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
networks:
- drone
ports:
- '80:80'
volumes:
- 'drone_data:/data'
drone-runner-docker:
container_name: drone-runner-docker
image: drone/drone-runner-docker:1
restart: unless-stopped
environment:
- DRONE_RPC_HOST=drone.internal.mydomain.com
- DRONE_RPC_PROTO=https
- DRONE_RUNNER_NAME=runner1
- DRONE_UI_DISABLE=true
- DRONE_RPC_SECRET=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
networks:
- drone
ports:
- '3000:3000'
volumes:
- '/var/run/docker.sock:/var/run/docker.sock'
networks:
drone:
volumes:
drone_data:
driver: local
Once you startup your containers and try to login to Drone (my instance is behind a reverse proxy), you should be greeted with your VCS provider’s login page. Once you login, you should be able to click Sync to sync your repositories to Drone.
I’m going to assume you already have an ansible repository, so click on it, and click Activate.
Configuration
Next, we need to create a configuration file for Drone. This file is usually called .drone.yml, and it’s right in the root of your git repository.
This configuration file needs to point to a Docker image that already has ansible
and ansible-lint
pre-installed. Luckily, Jeff Geerling already hosts an image on Docker Hub for this exact purpose (specifically, we need the image with the testing tag).
---
kind: pipeline
type: docker
name: default
steps:
- name: test
image: geerlingguy/docker-ubuntu1804-ansible:testing
commands:
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-playbook --syntax-check --list-tasks"
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-lint"
The configuration file above basically says:
- Use a Docker pipeline (so all these steps happen in Docker containers that go away after the run is completed)
- Run a step named test
- Download an image called geerlingguy/docker-ubuntu1804-ansible:testing
- Run the two commands for syntax checking and linting inside that container
- If there are any errors (non-zero exit codes), throw an error
- Based on the output (pass/fail), you’ll see a symbol in your VCS provider’s web interface (usually a ✅ or ❌ symbol).
Now, every time I commit anything to my Ansible repository, it is automatically checked for syntax and linted. The only remaining step (which I have yet to setup) is alerting. Drone has plugins to notify you via email, Slack, Telegram, etc…
Bonus: Run you own Ansible test images
Since I already run a Docker registry at home and have a Gitea server, I build my own Ubuntu images to test against. In fact, the image for Ubuntu is built with Drone as well!
FROM ubuntu:bionic
ARG BUILD_DATE
LABEL \
maintainer="Logan Marchione" \
org.opencontainers.image.authors="Logan Marchione" \
org.opencontainers.image.title="docker-ubuntu1804-ansible" \
org.opencontainers.image.description="Installs Ansible for testing playbooks" \
org.opencontainers.image.created=$BUILD_DATE
RUN apt-get update && apt-get install -y --no-install-recommends \
apt-utils \
python3-pip \
python3-setuptools \
software-properties-common && \
rm -rf /var/lib/apt/lists/* && \
pip3 install ansible ansible-lint
In my case, I’m able to reference my local internal registry instead of Docker Hub. I can also test against more than one version of Linux, or in my case, Ubuntu.
---
kind: pipeline
type: docker
name: default
steps:
- name: test-ubuntu1804
image: registry.internal.mydomain.com/loganmarchione/docker-ansible-ubuntu1804:1.0
commands:
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-playbook --syntax-check --list-tasks"
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-lint"
- name: test-ubuntu2004
image: registry.internal.mydomain.com/loganmarchione/docker-ansible-ubuntu2004:1.0
commands:
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-playbook --syntax-check --list-tasks"
- "find . -maxdepth 1 -name '*.yml' | sort | grep -v '.drone.yml' | xargs ansible-lint"
Conclusion
I hope this was a good introduction to Ansible linting and Drone. As a mentioned, I’m not an expert, so if you spot anything wrong or see anything I could improve, I’m open to criticism.
Thanks for reading!
-Logan