Pebble Mine

Setup a seed node for Radicle's peer-to-peer based source control on a fresh install of Ubuntu for a Raspberry Pi. This uses Ansible to provision your Raspberry.

Setup

Adjust the host variables in inventory.yml to suit your needs. In order for people to connect to the seed node from the internet, you'll have to allow incomming connections for the following ports:

• UDP:12345 - for peer data exchange
• TCP:80 - for the seed node UI

If you want to have a domain set for your seed you can add: --public-addr "{{ public_addr }}:12345" to the 'run seed node' task in roles/pebble/tasks/main.yml.

For example:

...
- name: run seed node
  shell: |
    cargo run -p radicle-seed-node --release -- \
    --root /home/{{ pi_username }}/{{ radicle_seed }} \
    --peer-listen 0.0.0.0:12345 \
    --http-listen 0.0.0.0:80 \
    --name "{{ radicle_hostname }}" \
    --public-addr "{{ public_addr }}:12345" \
    --assets-path seed/ui/public \
    < /home/{{ pi_username }}/{{ radicle_seed }}/secret.key
...

In user-data, change the values for <HOSTNAME>, <PEBBLE_PASSWD>, <PEBBLE_SSH_KEY>, and <PEBBLE_REPO>. You will need to create a new ssh key.

You can find the main config for the seed in roles/pebble/tasks/main.yml.

Steps

First you'll need a new instance of Ubuntu installed on an SD for the Raspiberry Pi, I tested this with v20.04.3 LTS. You can use any way you like to install it, I used RPI Imager.

Once the image has been copied, mount the drive and copy the user-data config file from the repo into the boot partition of SD card.

Boot the Raspberry Pi. You will need the secret key from when you created a key during the setup. It will take ≈10-15 minutes to complete the setup.

You will find the provisioning and troubleshooting logs at /var/log/ansible-provision.run and /var/log/cloud-init-output.log.

After the seed node is finished installing you can access it at <SEED_ID>>@<PUBLIC_ADDR>:<PORT>. If you didn't set a public_addr you can find it at your Raspberry's IP.

Note

This isn't totally finished... some things left to do:

• fix the reboot task to continue with the playbook after reboot
• add optional step for adding an LVM for the node