Aggravator

Dynamic inventory script for Ansible that aggregates information from other sources

Installing

virtualenv aggravator
source aggravator/bin/activate
pip install aggravator

Executing

ansible-playbook -i aggravator/bin/inventory site.yml

How does it work

It will aggregate other config sources (YAML or JSON format) into a single config stream.

The sources can be files or urls (to either file or webservices that produce YAML or JSON) and the key path to merge them under can be specified.

Why does it exist

We wanted to maintain our Ansible inventory in GIT as YAML files, and not in the INI like format that Ansible generally supports for flat file inventory.

Additionally we had some legacy config management systems that contained some information about our systems that we wanted exported to Ansible so we didn't have to maintain them in multiple places.

So a script that could take YAML files and render them in a JSON format that Ansible would ingest was needed, as was one that could aggregate many files and streams.

Config format

Example (etc/config.yaml):

---
environments:
  test:
    include:
      - path: inventory/test.yaml
      - path: vars/global.yaml
        key: all/vars
      - path: secrets/test.yaml
        key: all/vars

By default the inventory script will look for the root config file as follows:

• ../etc/config.yaml (relative to the inventory file)
• /etc/aggravator/config.yaml
• /usr/local/etc/aggravator/config.yaml

If it can't find it in one of those locations, you will need to use the --uri option to specify it (or set the INVENTORY_URI env var)

It will parse it for a list of environments (test, prod, qa, etc) and for a list of includes. The include section should be a list of dictionaries with the following keys:

path

The path to the data to be ingested, this can be one of: - absolute file path - relative file path (relative to the root config.yaml) - url to a file or service that emits a supported format

key

The key where the data should be merged into, if none is specified it is imported into the root of the data structure.

format

The data type of the stream to ingest (ie. yaml or json) if not specified then the script will attempt to guess it from the file extension

Order is important as items lower in the list will take precedence over ones specified earlier in the list.

Merging

Dictionaries will be merged, and lists will be replaced. So if a property at the same level in two source streams of the same name are dictionaries their contents will be merged. If they are lists, the later one will replace the earlier.

If the data type of two properties at the same level are different the later one will overwrite the earlier.

Environment Variables

Setting the following environment variables can influence how the script executes when it is called by Ansible.

INVENTORY_ENV

Specify the environment name to merge inventory for as defined under the 'environments' section in the root config. The environment name can also be guessed from the executable name, so if you create a symlink from prod to the inventory bin, it will assume the env you want to execute for is called prod, unless you override that.

INVENTORY_FORMAT

Format to output in, defaults to YAML in >0.4 Previously only output in JSON

INVENTORY_URI

Location to the root config, if not in one of the standard locations

VAULT_PASSWORD_FILE

Location of the vault password file if not in the default location of ~/.vault_pass.txt, can be set to /dev/null to disable decryption of secrets.

Usage

inventory [OPTIONS]

Ansible file based dynamic inventory script

Options:

--env TEXT	specify the platform name to pull inventory for
--uri TEXT	specify the URI to query for inventory config file, supports file:// and http(s):// [default: /home/peterb-l/git/petercb/aggravator/venv/etc/config.yaml]

--output-format [yaml|json] specify the output format [default: yaml] --vault-password-file PATH vault password file, if set to /dev/null secret

decryption will be disabled [default: ~/.vault_pass.txt]

--list	Print inventory information as a JSON object
--host TEXT	Retrieve host variables (not implemented)
--createlinks DIRECTORY
	Create symlinks in DIRECTORY to the script for each platform name retrieved
--show	Output a list of upstream environments (or groups if environment was set)
--help	Show this message and exit.