This repository contains the Agent Integrations (also known as checks) that Datadog officially develops and supports. To add a new integration, please see the Integrations Extras repository and the accompanying documentation.

The Datadog Agent packages are equipped with all the checks from this repository, so to get started using them, you can simply install the Agent for your operating system.

General documentation about the project can be found here

When working with an integration, you will now be dealing with a more structured python project. The new structure should help keep a more sane and modular codebase. To help with the transition, please take a look at the following map to understand where everything falls into place in the new approach.

Now that integrations are cleanly defined as python packages, we will soon be able to ship them as Python wheels that will be pip-installable in the Python environment embedded into the Datadog Agent. This presents a paradigm change in the way we will be delivering standalone integration upgrades, moving away from OS-specific packages to idiomatic Python package delivery.

Agent releases starting from version 5.21 bundle the latest wheels for any integration, but at the moment you can't upgrade or downgrade between releases.

You can follow these instructions to get a working copy of any check on your local Python environment; this is mostly useful to run tests or for tinkering in general.

• Python 2.7, see this page for more details.

The project comes with a requirements file you can pass to pip to install all the dependencies needed to work with any check. From the root of the repo, run:

pip install -r requirements-dev.txt

To work with a specific check you need to install its own dependencies. The easiest way to iterate on a check development is installing the wheel itself in editable mode. For example, if you want to do this for the disk check run the following:

cd disk && pip install -e .

To double check everything is working as expected you can run:

python -c"from datadog_checks.disk import Disk"

if the commands ends without errors, you're good to go!

To run the testsuite for a given check you can either use tox, like:

cd {integration} && tox

or invoke Pytest directly:

cd {integration} && py.test

Note: only a subset of the checks can be tested like this. Porting all the checks to Pytest is a work in progress, this is the list of the checks supporting the new testing approach:

• apache
• btrfs
• directory
• disk
• envoy
• istio
• kube_proxy
• kubelet
• lighttpd
• linkerd
• mcache
• network
• nfsstat
• postgres
• powerdns_recursor
• prometheus
• redisdb
• spark
• ssh_check
• system_core
• teamcity
• vsphere

For checks that are not listed here, please refer to Legacy development Setup.

If you updated the test requirements for a check, run tox --recreate for changes to be effective.

setup.py provides the setuptools setup script that will help us package and build the wheel. If you wish to learn more about python packaging please take a look at the official python documentation here

Once your setup.py is ready, creating a wheel is a easy as:

cd {integration}
python setup.py bdist_wheel

For more information on integrations, please reference our documentation and knowledge base. You can also visit our help page to connect with us.