Understand workspaces and deployment roles

Before you start: complete Compose a module and inspect its manifest, or make sure you can pass its checkpoint.

Read the application boundary#

A module contributes capabilities. A deployment selects the complete set of modules that run together. A workspace lists those deployments. The lab exports a deployments array from ket.workspace.ts; the CLI resolves the emitted JavaScript after the TypeScript build.

Open the lab workspace. modules: [learn_api] selects the module. headless: true means this deployment serves its API without selecting a theme. serve.routes adds its HTTP facade. The worker role later consumes jobs from the same composition.

Inspect selection#

# Run from: learn_api
npm run build
npx ket workspace --workspace dist/ket.workspace.js
npx ket manifest --deployment learn_api --workspace dist/ket.workspace.js

Use --deployment when a workspace contains more than one application. A feature name is not a deployment selection. Before running tests, identify the actual application composition those tests need.

Separate processes from contracts#

The HTTP process receives requests; the worker process claims background jobs. They can share a datastore while having different process roles. Both must load the same relevant module declarations, otherwise one role may enqueue a job that the other cannot execute.

A shared datastore across multiple deployments uses a checked union schema. Removing a model from one deployment does not prove no other deployment needs it. Treat composition and datastore selection as release decisions.

Follow environment configuration#

Keep secrets and host-specific locations out of module source. Use the documented runtime configuration for ports, database locations, signing keys and storage. Inspect the printed runtime banner when a process starts so you know which deployment and datastore you actually opened.

For the lab, use a private local data directory and never point a learning exercise at production credentials. A reset exercise should only recreate its disposable lab data.

Checkpoint#

Identify the lab deployment, module list, HTTP role and datastore. Explain why a worker needs the same job definitions as the producer.

Practice on your own#

Draw the HTTP and worker processes sharing a lab datastore. Mark which source file selects the modules and which settings vary by environment.

Reference#

For the complete API contract, read Workspaces.