Note: The Docker Daemon check is still maintained but only works with Agent v5.

Overview

Configure this Agent check to get metrics from the Docker_daemon service in real time to:

• Visualize and monitor Docker_daemon states.
• Be notified about Docker_daemon failovers and events.

Setup

Installation

To collect Docker metrics about all your containers, run one Datadog Agent on every host. There are two ways to run the Agent: directly on each host, or within a docker-dd-agent container (recommended).

For either option, your hosts need cgroup memory management enabled for the Docker check to succeed. See the docker-dd-agent repository for how to enable it.

Host installation

1. Ensure Docker is running on the host.
2. Install the Agent as described in the Agent installation instructions for your host OS.
3. Enable the Docker integration tile in the application.
4. Add the Agent user to the Docker group: usermod -a -G docker dd-agent
5. Create a docker_daemon.yaml file by copying the example file in the Agent conf.d directory. If you have a standard install of Docker on your host, there shouldn’t be anything you need to change to get the integration to work.
6. To enable other integrations, use docker ps to identify the ports used by the corresponding applications.

Container installation

1. Ensure Docker is running on the host.

2. As per the Docker container installation instructions, run:

    docker run -d --name dd-agent \
      -v /var/run/docker.sock:/var/run/docker.sock:ro \
      -v /proc/:/host/proc/:ro \
      -v /sys/fs/cgroup/:/host/sys/fs/cgroup:ro \
      -e API_KEY={YOUR_DD_API_KEY} \
      datadog/docker-dd-agent:latest
   

In the command above, you are able to pass your API key to the Datadog Agent using Docker’s -e environment variable flag. Other variables include:

Note: Add --restart=unless-stopped if you want your agent to be resistant to restarts.

Running the Agent container on Amazon Linux

To run the Datadog Agent container on Amazon Linux, make this change to the cgroup volume mount location:

docker run -d --name dd-agent \
  -v /var/run/docker.sock:/var/run/docker.sock:ro \
  -v /proc/:/host/proc/:ro \
  -v /cgroup/:/host/sys/fs/cgroup:ro \
  -e API_KEY={YOUR API KEY} \
  datadog/docker-dd-agent:latest

Alpine Linux based container

The standard Docker image is based on Debian Linux, but as of Datadog Agent v5.7, there is an Alpine Linux based image. The Alpine Linux image is considerably smaller in size than the traditional Debian-based image. It also inherits Alpine’s security-oriented design.

To use the Alpine Linux image, append -alpine to the version tag. For example:

docker run -d --name dd-agent \
  -v /var/run/docker.sock:/var/run/docker.sock:ro \
  -v /proc/:/host/proc/:ro \
  -v /sys/fs/cgroup/:/host/sys/fs/cgroup:ro \
  -e API_KEY={YOUR API KEY} \
  datadog/docker-dd-agent:latest-alpine

Image versioning

Starting with version 5.5.0 of the Datadog Agent, the Docker image follows a new versioning pattern. This allows Datadog to release changes to the Docker image of the Datadog Agent but with the same version of the Agent.

The Docker image version has the following pattern: X.Y.Z where X is the major version of the Docker image, Y is the minor version, Z represents the Agent version.

For example, the first version of the Docker image that bundles the Datadog Agent 5.5.0 is: 10.0.550

Custom containers and additional information

For more information about building custom Docker containers with the Datadog Agent, the Alpine Linux based image, versioning, and more, reference the docker-dd-agent project on Github.

Validation

Run the Agent’s status subcommand and look for docker_daemon under the Checks section.

Agent v6

The latest Docker check is named docker and written in Go to take advantage of the new internal architecture. Starting from v6.0, the Agent doesn’t load the docker_daemon check anymore, even if it is still available and maintained for Agent v5. All features are ported on version >6.0 , except the following deprecations:

• The url, api_version and tags* options are deprecated. Direct use of the standard Docker environment variables is encouraged.
• The ecs_tags, performance_tags and container_tags options are deprecated. Every relevant tag is collected by default.
• The collect_container_count option to enable the docker.container.count metric is not supported. docker.containers.running and .stopped should be used.

Some options have moved from docker_daemon.yaml to the main datadog.yaml:

• collect_labels_as_tags has been renamed docker_labels_as_tags and supports high cardinality tags. See the details in datadog.yaml.example.
• exclude and include lists have been renamed ac_include and ac_exclude. To make filtering consistent across all components of the Agent, filtering on arbitrary tags has been dropped. The only supported filtering tags are image (image name) and name (container name). Regexp filtering is still available, see datadog.yaml.example for examples.
• The docker_root option has been split in two options: container_cgroup_root and container_proc_root.
• exclude_pause_container has been added to exclude paused containers on Kubernetes and Openshift (defaults to true). This avoids removing them from the exclude list by error.

Additional changes:

• The TAGS environment variable was renamed to DD_TAGS.
• The Docker Hub repository has changed from datadog/docker-dd-agent to datadog/agent.

The import command converts the old docker_daemon.yaml to the new docker.yaml. The command also moves needed settings from docker_daemon.yaml to datadog.yaml.

Data Collected

Metrics

Events

The Docker integration produces the following events:

• Delete Image
• Die
• Error
• Fail
• Kill
• Out of memory (oom)
• Pause
• Restart container
• Restart Daemon
• Update

Service Checks

docker.service_up
Returns CRITICAL if the Agent is unable to collect the list of containers from the Docker daemon. Returns OK otherwise.
Statuses: ok, critical

docker.container_health
Returns CRITICAL if a container is unhealthy. Returns OK otherwise or UNKNOWN if the health is unknown.
Statuses: ok, critical, unknown

docker.exit
Returns CRITICAL if a container exited with a non-zero exit code. Returns OK otherwise.
Statuses: ok, critical

Note: To use docker.exit, add collect_exit_codes: true in your Docker YAML file and restart the Agent.

Troubleshooting

Need help? Contact Datadog support.

Further Reading

• Compose and the Datadog Agent
• DogStatsD and Docker
• The Docker Monitoring Problem (series)
• How to Monitor Docker Resource Metrics
• How to Collect Docker Metrics
• 8 Surprising Facts about Real Docker Adoption
• Monitor Docker on AWS ECS
• Dockerize Datadog
• Monitor Docker with Datadog