Iterate then deploy¶

This tutorial demonstrates the development-to-production workflow where you iterate on a workload with a draft artifact, then register it and create a governed deployment.

Workflow overview¶

Step 1: Create a development workload¶

Create a workload with an inline artifact. The artifact is created with status=draft by default, allowing iteration.

Step 2: Iterate on the artifact¶

During development, update the artifact and restart the workload to test changes.

2a. Update artifact (partial update with PATCH)¶

Use PATCH to update specific metadata fields (for example, name or description) while keeping others unchanged.

2b. Replace entire artifact (full update with PUT)¶

Use PUT to replace the full artifact definition, including the container spec. For changes to the container specification (image, entrypoint, probes, and so on), PUT is required; PATCH does not support full artifact updates.

2c. Restart workload to apply changes¶

# Stop the workload
curl -X POST "${DATAROBOT_ENDPOINT}/console/workloads/${WORKLOAD_ID}/stop" \
  -H "Authorization: Bearer ${DATAROBOT_API_TOKEN}"

# Wait a few seconds, then start
curl -X POST "${DATAROBOT_ENDPOINT}/console/workloads/${WORKLOAD_ID}/start" \
  -H "Authorization: Bearer ${DATAROBOT_API_TOKEN}"

Step 3: Lock artifact for production¶

When ready for production, change the artifact status from draft to registered:

curl -X PATCH "${DATAROBOT_ENDPOINT}/registry/artifacts/${ARTIFACT_ID}" \
  -H "Authorization: Bearer ${DATAROBOT_API_TOKEN}" \
  -H "Content-Type: application/json" \
  -d '{"status": "registered"}'

Step 4: Create production deployment¶

Create a governed deployment using the WORKLOAD_ID from Step 1.

Key concepts¶