What is this for?

This tool/library exists to:

• provide a straightforward way to specify a particular set of DigitalOcean droplets and, given that specification,

• create/destroy them in an idempotent manner and

• provide an Ansible inventory for those droplets.

The work was motivated by the “chicken/egg” problem of using a configuration management tool like Ansible on a yet-uncreated DigitalOcean droplet/service cluster. Ansible is certainly useful for deploying a configuration to an existing set of resources, and it even has an API for creating those resources on DigitalOcean.

However, there wasn’t a straightforward way of using Ansible to both create and configure the resources without having an inventory file hand-written prior to creation. If you changed the number of created droplets, you would have to hand-edit the inventory. With this tool, an existing set of Ansible roles can be used on a dynamic set of DigitalOcean droplets.

dowright is for Python 3.4+ only with no plans for backwards compatibility.

Installation

• Install using pip3:

  pip3 install dowright

• Initialize the .tokenmanager.yml file in your home directory:

  python3 -m tokenmanager -i

• Provide a digitalocean token namespace in your .tokenmanager.yml file. Then provide as many sub-namespaces as you need for your YAML specifications. These will then be specified in your YAML using the token key. E.g.

  digitalocean:
    app1: a1e1c084540b51b33af3c6b63d48ede2937c8df92f7e6e3beb1f630ac750b851
    app2: 03593464105708646cc04d847ffc81c5b7775c462f68b573f2aff5d933635e17

Usage

To use dowright, you will need to create a YAML specification for your DigitalOcean resources. Here’s an example spec (named my_web_app.yml):

# token: this specifies which token reference under the
#   "digitalocean" token group should be used for creating
#   the droplets specified
token: app1 # using the above example of a .tokenmanager.yml file

# prefix: is used to namespace your resources and is also
#   provided as a tag for these resources at creation time
prefix: my_web_app

# droplets: is a dictionary of named groups of resources. Each group
#   corresponds to an Ansible inventory group and can be referenced
#   as a group in various ways in this specification.
droplets:
  # Groups are lists of dictionaries that correspond to the creation
  #   parameters of the droplets to be created. Group names are also
  #   added as tags at creation time.
  data:
    - name: data[01:10]  # can use Ansible-style name expansions
                         # to create multiple droplets with the same
                         # name prefix and parameters
      size_slug: 2gb  # each key in the dictionary corresponds 1-to-1
                      # with the parameters given in the DigitalOcean
                      # droplet creation REST API
      volumes:
        - aaaaaaaa-bbbb-ffff-3333-000000000000

      # cloud_config_commands: is a special key that will be
      #   transformed into the "#cloud-config" YAML-formatted string
      #   under the "user_data" parameter. It is a list of bash shell
      #   commands that will be placed in the "runcmd" list of the
      #   "#cloud-config" string. In order to track completion of
      #   these commands, a final command (that creates a sentinel
      #   file "/.cloud-config-done") is appended by dowright.
      cloud_config_commands:
        - mkdir /data
        - mount -o defaults,nofail /dev/disk/by-id/scsi-0DO_Volume_volume-nyc1-01-part1 /data
        - >-
          echo "/dev/disk/by-id/scsi-0DO_Volume_volume-nyc1-01-part1 /data xfs defaults,nofail 0 2" >> /etc/fstab
        - apt update
        - apt install -y python3 python3-pip
        - pip3 install --upgrade pip
  nameserver:
    - name: ns
      image: ubuntu-16-04-x64

# defaults: these are default creation parameters for all resources
defaults:
    image: ubuntu-17-04-x64
    size_slug: 1gb
    region: nyc1
    ssh_keys:
      - 999999
      - 999998
    private_networking: yes

# floating_ips: these are mappings of previously-created floating IPs
#   to be mapped to particular droplets. *Notice*: the name given here
#   corresponds to that given under the above "droplets:" section.
#   It does /not/ have the "prefix:" string (e.g. "my_web_app"
#   given above).
floating_ips:
  192.16.1.1: data01
  192.16.1.2: ns

# domains: mappings of DigitalOcean-managed domains to a list of
#   creation parameters for subdomains
domains:
  mydomin.com:
    - type: 'A'
      name: 'ds_master'
      data: data01
    - type: 'A'
      name: 'ns_master'
      data: ns

# inventory: this defines the Ansible inventory for the
#   DigitalOcean droplets.
inventory:
  # name: filename of inventory
  name: hosts.conf
  # groups: group definitions in inventory
  groups:
    # each group is a list of references to DigitalOcean droplets
    "datanodes:children":
      - name: data   # references can be to droplet groups
    namenodes:
      - name: nameserver
    hadoop_startup:
      - name: ns     # references can be to individual droplet names
    "nodemanagers:children":
      - name: datanodes  # if the inventory group name has a colon
                         #   in it (i.e. it's a group of inventory
                         #   groups), then the reference must be to
                         #   another inventory group
    "hadoop:children":
      - name: namenodes
      - name: datanodes
    analysis:
      - name: data01

To create/build the droplets specified:

python3 -m dowright my_web_app.yml -b

To wait for completion of droplet creation:

python3 -m dowright my_web_app.yml -w

To link floating IPs to DigitalOcean droplets:

python3 -m dowright my_web_app.yml -I

To link DigitalOcean-managed domains to droplets:

python3 -m dowright my_web_app.yml -d

To create the Ansible inventory for your droplets:

python3 -m dowright my_web_app.yml -i

To do all the above:

python3 -m dowright my_web_app.yml -bwiId

To destroy the DigitalOcean droplets specified in your YAML:

python3 -m dowright my_web_app.yml --destroy