Overview of Cartographer Conventions

This topic gives you an overview of Cartographer Conventions and how you can use it with Tanzu Application Platform (commonly known as TAP).

Overview

Cartographer Conventions provides a means for operators to express their knowledge about how applications can run on Kubernetes as a convention. Cartographer Conventions supports defining and applying conventions to pods. It applies these opinions to fleets of developer workloads as they are deployed to the platform, saving operator and developer time.

The service has two components:

• The convention controller:

  The convention controller provides the metadata to the convention server and runs the updates to a PodTemplateSpec in accordance with convention server’s requests.

• The convention server:

  The convention server receives and evaluates metadata associated with a workload and requests updates to the PodTemplateSpec associated with that workload. You can have one or more convention servers for a single controller instance.

About applying conventions

The convention server uses criteria defined in the convention to detect whether the configuration of a workload must change. The server receives the OCI metadata from the convention controller. If the metadata meets the criteria defined by the convention server, the conventions are applied. A convention can apply to all workloads regardless of metadata.

Applying conventions by using image metadata

You can define conventions to target workloads by using properties of their OCI metadata.

Conventions can use this information to only apply changes to the configuration of workloads when they match specific criteria. Such as, Spring Boot or .Net apps, or Spring Boot v2.3 and later. Targeted conventions can ensure that uniformity across specific workload types deployed on the cluster.

You can use all the metadata details of an image when evaluating workloads. To see the metadata details, use the Docker CLI command:

docker image inspect IMAGE.

Note

Depending on how the image was built, metadata might not be available to reliably identify the image type and match the criteria for a convention server. Images built with Cloud Native Buildpacks reliably include rich descriptive metadata. Images built by some other process might not include the same metadata.

Applying conventions without using image metadata

Conventions can apply to workloads without targeting build service metadata. Examples of possible uses of this type of convention include:

• Appending a logging or metrics sidecar
• Adding environment variables
• Adding cached volumes

These kinds of conventions ensure that infrastructure uniformity exists across workloads deployed on the cluster while reducing developer toil.

Important

Adding a sidecar alone does not make the log or metrics collection work. To enable the log or metrics collection:

• Deploy collector agents that are accessible from the Kubernetes cluster
• Configure the required access by using the role-based access control (RBAC) policy