There are three ways to monitor containers using Dynatrace. The method you choose depends on the following:
- The type of container runtime, such as Docker, containerd, or CRI-O
- The orchestration platform, such as Kubernetes, OpenShift, Cloud Foundry, and Fargate
- The level of access you have to the underlying host
See below for details.
Full stack injection
The most comprehensive option for monitoring containers with Dynatrace is to deploy OneAgent to your container platform. This assumes full access to the underlying host.
- If you're using the Docker runtime without an orchestration platform, see Set up Dynatrace OneAgent as a Docker container
- For Kubernetes, see Set up Kubernetes monitoring
- For OpenShift, see Set up OpenShift monitoring
- For Cloud Foundry, see Deploy OneAgent BOSH release on Cloud Foundry
Use application-only injection if you don't have access to underlying hosts. Your options depend on the container platform you are using.
The most efficient option is automated application-only injection for Kubernetes-based platforms. This injects OneAgent using Kubernetes-native admission controllers.
- For Kubernetes, see Kubernetes automatic application-only injection
- For OpenShift, see OpenShift automatic application-only injection
Runtime injection puts OneAgent inside a container as it is deployed.
- For Kubernetes, see Kubernetes pod runtime application-only injection
- For OpenShift, see OpenShift pod runtime application-only injection
- For Cloud Foundry, see Deploy OneAgent on IBM Cloud Foundry for application-only monitoring
All platforms also support the build-time approach, which puts OneAgent inside a container as it builds.
- For Docker, see Set up OneAgent on containers for application-only monitoring
- For Kubernetes, see Kubernetes container build-time application-only injection
- For OpenShift, see OpenShift container build-time application-only injection