About

ARC is implemented as a Kubernetes API Extension Server that provides an API for managing orders. As a result, ARC Order resources can be managed using kubectl and natively integrate with other Kubernetes services.

This walk-through builds a small but real use case step by step: transferring a file from an HTTP source into an S3 bucket. Each step adds one concept on top of the previous one. Progress through the walk-through in sequence to learn all the basics.

1. Setup: a local cluster with ARC, Argo Workflows and an S3 store.
2. First transfer: the minimal set of resources to move a blob from HTTP to S3.
3. Verify before you push: a multi-step workflow that checks a checksum and scans for viruses.
4. Conditionals: pull from HTTP or S3, depending on the source of each artifact.
5. Schedule transfers: transfer an artifact again and again with a cron schedule.
6. Next steps: more examples, Argo features and running ARC in production.

ARC and Argo Workflows

ARC itself does not move a single byte. It is a smart scheduler for Argo Workflows:

• ARC knows what to transfer (Order), from where to where (Endpoint) and how (ArtifactType).
• ARC validates orders, resolves credentials and creates one Argo Workflow per artifact with the right parameters and secrets wired in.
• The actual work (pulling, verifying, scanning, pushing) happens in an Argo WorkflowTemplate that you write.

This means that to build your own artifact types, you will write Argo Workflows. The walk-through explains the parts ARC is responsible for and links to the Argo Workflows documentation for everything else. Getting familiar with Argo Workflows is well worth the time.

For a complete description of the ARC Order spec, please refer to the spec documentation. Many more examples are available in the examples directory on GitHub.