DigitalOcean Meta-Write

dometawrite is a tool that helps system administrators take information from the DigitalOcean API and/or Dropleta Metadata API to create text files such as configuration files, status pages or anything else that can be coded in a template.

Templates are written using the most-excellent Jinja2 library for Python.

Example: Create SSH client configuration

dometawrite can be used to create an OpenSSH SSH Client configuration file that can be included from ~/.ssh/config for easy access to your droplets. Picture this:

~ $ dometawrite --template ssh-config  \
                --api-key $DO_API_KEY  \
                --output /home/example/.ssh/digitalocean_droplets \
                -u user:remoteUser -u keyfile:/home/example/id_rsa_digitalocean

The above command will render the ssh-config dometawrite template, and write the output to the /home/example/.ssh/digitalocean_droplets file.

It will use the indicated DigitalOcean API KEY, and will also provide some additional variables to the template. If any variables are missing from the command line, dometawrite will let you know.

The resulting output may look like this:

Host droplet-name
HostName droplet_ip_address
User root
IdentityFile /home/example/id_rsa_digitalocean

Host another_droplet
HostName yet_another_ip
User root
IdentityFile /home/example/id_rsa_digitalocean

As long as the id_rsa_digitalocean.pub file has been added to the droplet, either during creation or afterwards, then you will be able to simply:

ssh root@another_droplet

Example: Create OpenSSH AuthorizedKeysFile-compatible output using from your DigitalOcean account.

This is a useful feature, but please do NOT --output directly to your authorized_keys file!

dometawrite --template authorizedkeys \
            --api-key $DO_API_KEY

Example: Create DROPLET_NAME, DROPLET_TAGS, DROPLET_REGION environment variables in bash

Run this in your droplet:

dometawrite -t droplet_envvars

You will get a bash-compatible output with a number of export commands.

If you are running a modern-enough Linux distribution with /dev/fd available, you can then add this line to your ~/.bashrc:

source <(dometawrite -t droplet_envvars)

You can use that in Ubuntu 20.04, maybe 19.x - But not in 18.04, where you would need to:

dometawrite -t droplet_envvars -o ~/.de.tmp
source ~/.de.tmp

From then on, when you access your droplet's bash shell you will have a couple additional variables at your disposal.

Example: Create ansible inventory (This template is not yet available)

Maybe you use ansible, and you want to update your hosts inventory dynamically:

dometawrite --template ansible-inventory \
            --api-key $DO_API_KEY        \
            --output /etc/ansible/hosts/digitaloceaninventory

The template will receive a python dictionary with all necessary information. Jinja2 supports advanced logic, so it can easily contain all the required code to output a valid Ansible Inventory file.

Templating Syntax

We use jinja2 for the templating support, so anything it supports, dometawrite supports. But we do have some specific requirements:

Templates will have to indicate endpoint requirements, and specific user variables.

For example, for the ssh-config template:

{% set endpoint_requirements = ['droplets'] %}
{% set userdata_requirements = ['user','keyfile'] %}

{% for droplet in droplets.droplets %}
Host {{ droplet.name }}
HostName {{ droplet.networks.v4[0].ip_address }}
User {{ userdata.user }}
IdentityFile {{ userdata.keyfile }}

{% endfor %}

Some endpoints include '/' in their name. The template-context variable name is automatically named replacing all instances of '/' with '_'. For example, take the authorizedkeys template:

{% set endpoint_requirements = ['account/keys'] %}
{% for key in account_keys.ssh_keys %}
# Key "{{ key.name }}" ID={{ key.id }} Fingerprint={{key.fingerprint}}
{{ key.public_key }}
{% endfor %}

You can see that we iterate over each key using account_keys.ssh_keys as source, but the endpoint_requirements is set to call 'accounts/keys' DO APIv2 endpoint.

Note

The syntax is bound to change a bit to support features (for example, author details support, template description, etc).

TODO

• add testing https://python-packaging.readthedocs.io/en/latest/testing.html
• add pagination support
• Think about --private-only andd/or --public-only modes? should be an arg, or a template-level function?