Installing the Datadog Agent to Collect Tests Data

To report test results to Datadog, the Datadog Agent is required.

There are two ways to set up the Agent in a CI environment:

Installing it as a long running process on each worker node.
Running an ephemeral instance of it as a service container on each build.

Installing the Agent on each CI worker node

If you are running tests on an on-premises CI provider, where test processes have network access to the underlying worker host, install the Datadog Agent on each worker node the Agent installation instructions.

Once the Agent is installed and running, the tracer can send test results to localhost:8126.

Installing the Datadog Agent as a service container on each build

If your CI provider runs your tests in a container (such as on-prem CI providers using a Docker or Kubernetes executor, or a SaaS CI provider), run the Datadog Agent in a container as a build service.

To enable APM and disable the monitoring of the container, set the following environment variable on the Agent container:

DD_INSIDE_CI=true
DD_HOSTNAME=none

This allows the tracer to send test results to port 8126 of the Agent container.

Running the Datadog Agent using Docker Compose

Regardless of which CI provider you use, if tests are run using Docker Compose, the Agent can be run as one service:

docker-compose.yml

version: '3'
services:
  datadog-agent:
    image: "datadog/agent:latest"
    environment:
      - DD_API_KEY
      - DD_HOSTNAME=none
      - DD_INSIDE_CI=true
    ports:
      - 8126/tcp

  tests:
    build: .
    environment:
      - DD_AGENT_HOST=datadog-agent

Alternatively, share the same network namespace between the Agent container and the tests container:

docker-compose.yml

version: '3'
services:
  datadog-agent:
    image: "datadog/agent:latest"
    environment:
      - DD_API_KEY
      - DD_HOSTNAME=none
      - DD_INSIDE_CI=true

  tests:
    build: .
    network_mode: "service:datadog-agent"

In this case, no DD_AGENT_HOST is required because it is localhost by default.

Then, run your tests by providing your Datadog API key in the DD_API_KEY environment variable:

DD_API_KEY=<MY_API_KEY> docker-compose up --build --abort-on-container-exit tests

Note: In this case, also pass along all the required CI provider environment variables so build information can be attached to each test result, as described in Tests in containers.

CI provider-specific instructions

The following sections provide CI provider-specific instructions to run and configure the Agent to report test information.

Azure Pipelines

To run the Agent in Azure Pipelines, define the Agent container in the resources section and link it with the job declaring it as a service container:

azure-pipeline.yml

variables:
  ddApiKey: $(DD_API_KEY)

resources:
  containers:
    - container: dd_agent
      image: datadog/agent:latest
      ports:
        - 8126:8126
        env:
          DD_API_KEY: $(ddApiKey)
          DD_HOSTNAME: "none"
          DD_INSIDE_CI: "true"

jobs:	
  - job: test
    services:
      dd_agent: dd_agent
    steps:
      - script: make test

Add your Datadog API key to your project environment variables with the key DD_API_KEY.

GitLab CI

To run the Agent in GitLab, define the Agent container under services:

.gitlab-ci.yml

variables:
  DD_AGENT_HOST: "datadog-agent"
  DD_HOSTNAME: "none"
  DD_INSIDE_CI: "true"
  DD_API_KEY: $DD_API_KEY

test:
  services:
    - name: datadog/agent:latest
  script:
    - make test

Add your Datadog API key to your project environment variables with the key DD_API_KEY.

GitHub Actions

To run the Agent in GitHub Actions, define the Agent container under services:

jobs:
  test:
    services:
      datadog-agent:
        image: datadog/agent:latest
        ports:
          - 8126:8126
        env:
          DD_API_KEY: ${{ secrets.DD_API_KEY }}
          DD_HOSTNAME: "none"
          DD_INSIDE_CI: "true"
    steps:
      - run: make test

Add your Datadog API key to your project secrets with the key DD_API_KEY.

CircleCI

To run the Agent in CircleCI, launch the agent container before running tests by using the datadog/agent CircleCI orb, and stop it after to ensure results are sent to Datadog.

For example:

.circleci/config.yml

version: 2.1

orbs:
  datadog-agent: datadog/agent@0.0.1

jobs:
  test:
    docker:
      - image: circleci/<language>:<version_tag>
    steps:
      - checkout
      - datadog-agent/setup
      - run: make test
      - datadog-agent/stop

workflows:
  test:
    jobs:
      - test