Datadog Agent 6+ collects logs from containers. Two types of installation are available:
Configuring log collection depends on your current environment. Choose one of the following installations to get started:
If your environment writes all logs to stdout/stderr, follow the containerized Agent installation.
If you cannot deploy the containerized Agent and your container writes all logs to stdout/stderr, follow the host Agent installation to enable containerized logging within your Agent configuration file.
The CLI commands on this page are for the Docker runtime. Replace docker with nerdctl for the containerd runtime, or podman for the Podman runtime. Support for containerd and Podman log collection is limited.
To prevent loss of container logs during restarts or network issues, the last log line collected for each container in this directory is stored on the host.
-e DD_CONTAINER_EXCLUDE="name:datadog-agent"
Prevents the Datadog Agent from collecting and sending its own logs and metrics. Remove this parameter if you want to collect the Datadog Agent logs or metrics. This parameter value supports regular expressions.
-v /var/run/docker.sock:/var/run/docker.sock:ro
To connect to the Docker daemon to discover containers and collect stdout/stderr from the Docker socket.
To collect containers logs from files. Available in the Datadog Agent 6.27.0/7.27.0+
Note: If using Docker Compose, the value for DD_CONTAINER_EXCLUDE must not be quoted. Configure the environment variable in your docker-compose.yaml file like the example below:
Windows 10 Only: The Datadog Agent user must be a member of the docker-users group in order to have permissions to work with Docker containers. Run net localgroup docker-users "ddagentuser" /ADD from your Administrator command prompt or follow the Docker User Group configuration steps.
To gather logs from your <APP_NAME> application stored in <PATH_LOG_FILE>/<LOG_FILE_NAME>.log create a <APP_NAME>.d/conf.yaml file at the root of your Agent’s configuration directory with the following content:
Note: In order for the Agent to collect logs produced by a container with a custom log configuration, the logs must be written to a volume accessible from the host. It is recommended that container logs be written to stdout and stderr so that they can be collected automatically.
Important notes:
Container metadata is not retrieved with custom log collection, therefore the Agent does not automatically assign container tags to logs. Use custom tags to create container tags.
source and service default to the short_image tag value in Datadog Agent 6.8+. The source and service values can be overridden with Autodiscovery as described below. Setting the source value to an integration name results in the installation of integration Pipelines that parse your logs and extract relevant information from them.
Logs coming from container Stderr have a default status of Error.
If using the journald logging driver instead of Docker’s default json-file logging driver, see the journald integration documentation for details regarding the setup for containerized environments. See the journald filter units documentation for more information on parameters for filtering.
In Datadog Agent 6.8+, source and service default to the short_image tag value. This allows Datadog to identify the log source for each container and automatically install the corresponding integration.
The container short image name might not match the integration name for custom images, and can be overwritten to better reflect the name of your application. This can be done with Datadog Autodiscovery and pod annotations in Kubernetes or container labels.
Autodiscovery expects labels to follow this format, depending on the file type:
Note: When configuring the service value through docker labels, Datadog recommends using unified service tagging as a best practice. Unified service tagging ties all Datadog telemetry together, including logs, through the use of three standard tags: env, service, and version. To learn how to configure your environment with unified tagging, see the unified service tagging documentation.
For multi-line logs like stack traces, the Agent has multi-line processing rules to aggregate lines into a single log.
Example log (Java stack traces):
2018-01-03T09:24:24.983Z UTC Exception in thread "main" java.lang.NullPointerException
at com.example.myproject.Book.getTitle(Book.java:16)
at com.example.myproject.Author.getBookTitles(Author.java:25)
at com.example.myproject.Bootstrap.main(Bootstrap.java:14)
Use the com.datadoghq.ad.logs label as below on your containers to make sure that the above log is properly collected:
The Agent v7.25.0+/6.25.0+ can directly collect logs from a file based on a container Autodiscovery label. To collect these logs, use the com.datadoghq.ad.logs label as shown below on your containers to collect /logs/app/prod.log:
Logs collected from a file are tagged with the container metadata. Log collection is linked to the container life cycle, as soon as the container stops, log collection from that file stops.
Notes:
The file path is relative to the Agent, so the directory containing the file should be shared between the container running the application and the Agent container. For example, if the container mounts /logs each container logging to file may mount a volume such as /logs/app where the log file is written.
When using this kind of label on a container, its stderr/stdout logs are not collected automatically. If collection from both stderr/stdout and a file are needed it should be explicitly enabled by using a label, for example:
When using this kind of combination, source and service have no default value and should be explicitly set in the Autodiscovery label.
Note: Autodiscovery features can be used with or without the DD_LOGS_CONFIG_CONTAINER_COLLECT_ALL environment variable. Choose one of the following options:
Use container labels or pod annotations to choose the containers to collect logs from.
Use the environment variable to collect logs from all containers and then override the default source and service values.
Add processing rules for the wanted subset of containers.
Docker container log collection from a file is an alternative to collection over the Docker socket. File based collection offers better performance than socket based collection.
In versions 7.27.0/6.27.0+, you can configure the Agent to collect Docker container logs from a file. In versions 6.33.0+/7.33.0+, the Agent collects Docker container logs from a file by default.
File based collection requires the directory storing Docker container logs to be exposed to the Agent in the following location: /var/lib/docker/containers (c:\programdata\docker\containers on Windows). See the Docker logs collection troubleshooting guide for more information.
Note:
When you migrate from Docker socket based container log collection to file based log collection, only new containers are tailed from their files. You can force the Agent to collect all container logs from files by setting the environment variable DD_LOGS_CONFIG_DOCKER_CONTAINER_FORCE_USE_FILE to true. Forcing the Agent to collect all container logs from files may result in duplicated logs for existing containers.
If you switch the Agent back from container file log collection to collection over the Docker socket, you will likely see duplicated logs for existing containers.
It is possible to manage from which containers you want to collect logs. This can be useful to prevent the collection of the Datadog Agent logs for instance. See the Container Discovery Management to learn more.
For a Docker environment, the Agent receives container updates in real time through Docker events. The Agent extracts and updates the configuration from the container labels (Autodiscovery) every 1 seconds.
Since Agent v6.14+, the Agent collects logs for all containers (running or stopped) which means that short lived containers logs that have started and stopped in the past second are still collected as long as they are not removed.
