Instrument your container with the infrastructure agent

Our infrastructure agent automatically monitors your containers. With container monitoring you can:

• Group containers by tags, attributes, and other metadata.
• Search for containers relevant to your monitoring scenario.
• Link to related entities that may be affected by issues with the container.
• Set container-related alert conditions.

Requirements

Requirement details for automatic container monitoring for New Relic's infrastructure agent:

• A New Relic account. Don't have one? Sign up for free!. No credit card required.
• Infrastructure agent 1.8.32 or higher running on Linux
• If using CentOS, you must have CentOS version 6.0 or higher
• Docker with engine from v1.12 or other containerd-based container runtimes.

Enable container monitoring

If you meet the requirements and have installed the correct infrastructure agent, there are no additional steps to enable container monitoring. If a container runtime is running, data will automatically be reported.

You can also use a container image containing the infrastructure agent. For more information, see Container for infrastructure monitoring.

View your container data

To view your container data in the New Relic UI, use either of these options:

• Go to one.newrelic.com > All capabilities > Infrastructure > Hosts > Containers.

  OR

• Go to one.newrelic.com > All capabilities > Infrastructure > Third-party services, and select Docker-related links.

For more information, see Query your data.

container attributes

Container attributes (metrics and metadata) are attached to the ContainerSample event. Here's an example of a query to find out how many containers are associated with each container image:

SELECT uniqueCount(containerId) FROM ContainerSample FACET imageName SINCE 1 HOUR AGO TIMESERIES

To see all ContainerSample attributes, use our data dictionary. Attributes include:

• General metadata (like containerId, name, and image)
• CPU metrics (like cpuUsedCores, cpuPercent, and cpuThrottleTimeMs)
• Memory metrics (like memoryUsageBytes, memoryCacheBytes, and memoryResidentSizeBytes)
• Network metrics (like networkRxBytes, networkRxDropped, and networkTxBytes)

Container metrics are also attached to the ProcessSample event.

The reported data does not include information related to the container orchestrator (for example, ECS or Kubernetes). To monitor those, you can add the orchestrator's cluster and task names as labels.

Set alert conditions

To create container-related alert conditions, use either of these options:

1. Go to one.newrelic.com > All capabilities > Alerts .

   OR

   Go to one.newrelic.com > All capabilities > Infrastructure > Settings > Alerts.

2. Create a new alert condition.

3. For the condition type, select Container metrics.

Enable container metrics collection from Docker API

The nri-docker integration, by default, employs the Docker API in conjunction with the /proc filesystem to extract container metrics. As of version v1.51.0 of the infrastructure agent, you can reconfigure the integration to solely source metrics from the Docker API.

This Docker API-only collection strategy applies only when the Docker Engine Cgroup operates under version V2. To enforce the Docker API-only collection, adapt the docker-config.yml configuration file located in the integrations.d directory to resemble the following:

integrations:
  - name: nri-docker
    env:
      USE_DOCKER_API: true
    when:
      feature: docker_enabled
      file_exists: /var/run/docker.sock
    interval: 15s

Disabling container monitoring

To disable this capability simply delete the docker-config.yml configuration file from the integrations.d folder.