Ansible playbook that manages a cluster of local Raspberry Pi machnes running Pihole and Time Machine.

This is a personal learning project, please don't use that in production ;) If you see that I did something very wrong here, please do file an issue or DM me, I'd be very grateful for an opportunity to learn more.

If you have a bunch of PIs and you'd like to play around with Ansible, do this:

1. Fork and clone the repo
2. Edit the inventory/hosts.ini file to include the list of your Raspberry PIs hostnames
3. Edit the group_vars/all.yaml, most importantly the dotfiles repo and github username used for ssh keys.

The roles will work for both Raspbian and Ubuntu 20.04, you can set your machines up with whatever system you prefer. I'm using Ubuntu for the docker host, and the K3s cluster, and Raspbian on Pi Zero machines.

1. Follow the docs on starting the Pi up. Preparing the OS microsd memory card and enabling wifi on it are not covered by this repo's playbooks.
2. Manually ssh into the new PI and setup it's hostname in /etc/hostname and /etc/hosts. I'm using mDNS and naming rpiX.local.
3. Add the new host to hosts inventory.
4. ansible-playbook site.yaml will run ssh-keys, prereq, pihole, and time-machine roles. If you just need pihole OR time machine, you can comment out the roles that you don't need in the site.yaml.

Assuming the card already have a fresh raspberrian lite on it:

cd /Volumes/boot # Assuming you're doing this in mac OS.
touch ssh        # To enable SSH by default on the new PI
vim wpa_supplicant.conf # Wifi network, if needed.

See more here.

Puts your ssh keys from github into the Pi, and disables password authentication in ssh.

Settings:

• github_username in group_vars/all.yaml — sets username in github to grab public keys from.

Sets up requirements for any additional roles on the PIs, and prepares the environment.

What it does:

• Set timezone to timezone var. Default: Pacific.
• Add github.com to known_hosts so PI can clone / checkout repos.
• Installs essential packages, for now it explicitly uses apt, not package module. You can setup packages var to install additional packages. By default, it installs git, fish, vim, and tmux.
• Updates all packages to latest versions.
• Check out and install dotfiles from a repo set in dotfiles_repo

Note: on Ubuntu, I use a cloud-init user-data script that manages the majority of those tasks, and one of the next steps for this repository is to have a separate playbook that burns an SD card and sets up all required files, including network-config and user-data for Ubuntu, and ssh and wpa_supplicant on Raspbian.

Installs and runs pihole in a docker container in all machines in pihole group.

Note: group_vars/docker.yaml is vaulted, and has pihole and time machine passwords in there. For your own cluster, you'll want to rm the file, and create your own vars file, optionally vaulting it, too.

The role uses geerlingguy.pip and geerlingguy.docker roles as requirements, and then uses docker-compose to start and manage pihole.

Settings:

• pihole_hostname — which hostname to set for pihole. That might be specific to your network.
• Take a look at pihole/defaults/main.yaml
  for more variables that you can set. They're all optional and have a sensible default value.

Runs mbentley/docker-timemachine in the cluster in docker-compose. Takes timemachine_password variable in, it's defined in the vaulted group_vars/dockerized.yaml by default.

Settings:

• Take a look at roles/time-machine/defaults/main.yaml to learn more. Hostname, password, and volumes are customizable.

By default, time machine role uses internal hfsplus role to mount an HFS+ (Mac OS Extended) partition on an external drive /dev/sda2. If you'd rather have an Ext4 drive, you could drop the HFS+ dependency.

This playbook is runnning apps with Docker Compose. All of the setting and data of the apps are in mounted volumes, and volumes, by default, are in /home/pi/volumes/. You can change this path by changing docker_volumes_path variable in group_vars/all.

docker-compose.yaml files for each app are stored in /home/pi/apps/*.

Want to help out? Take a look at TODOs