Deploy pipeline bundles from local machine¶
Deploy a Pipeline Bundle you authored to your Databricks workspace from your local machine using the Declarative Automation Bundle (DAB) CLI workflow.
The Framework Bundle must already be deployed in the target workspace — central/platform path or your own dev deploy. See Before you deploy for deploy order.
For deploy order and ownership models, see Before you deploy. For automated pipelines, see Setting up CI/CD.
Prerequisites¶
Before you begin, verify:
Databricks workspace access with permission to deploy bundles and run Lakeflow Spark Declarative Pipelines
Databricks CLI installed — required for local Asset Bundle deployment (CLI documentation)
CLI authentication — run
databricks auth loginfor your workspace, or use a configured CLI profileUnity Catalog enabled in your workspace
Framework deployed at a known
framework_source_pathin this workspace
Step 1 — Authenticate the Databricks CLI¶
Authenticate against the workspace you will deploy to:
databricks auth login --host https://<your-workspace-url>
Or use an existing named profile. Pass the profile on later commands with -p <profile>:
databricks bundle deploy -t dev -p <profile>
Step 2 — Configure the workspace host (optional)¶
Ensure the correct Databricks workspace is selected. Either:
Leave
workspace.hostunset indatabricks.ymlso the CLI uses the host from the selected profile, orSet the host explicitly under the target you will deploy to.
The Databricks CLI must be authenticated with credentials that can access this workspace.
Step 3 — Set the framework source path¶
Point your pipeline bundle at the deployed framework version in workspace files. Set framework_source_path in databricks.yml (or pass --var), for example:
/Workspace/Users/<owner>/.bundle/lakeflow_framework/dev/current/files/src
If you deployed the framework yourself for local dev, the default path is under your user .bundle/.../files/src. See Deploy framework from local machine.
Step 4 — Validate the bundle¶
From your pipeline bundle directory (after Building a Pipeline Bundle):
databricks bundle validate
This runs checks to ensure the bundle is correctly set up and ready for deployment.
Step 5 — Deploy the bundle¶
databricks bundle deploy -t dev
When the framework path is not already set in databricks.yml, pass it at deploy time:
databricks bundle deploy -t dev --var="framework_source_path=/Workspace/Users/<owner>/.bundle/lakeflow_framework/dev/current/files/src"
Note
By default the CLI deploys to the dev target in databricks.yml.
Use -t <target> to deploy elsewhere and -p <profile> to select a different CLI profile.
When pinning to a specific framework version for rollback, see Versioning - Framework.
Step 6 — Verify the deployment¶
When deployment succeeds, bundle files are present in your Databricks workspace under the .bundle directory for the deploying user (see workspace.root_path in databricks.yml).
Open the workspace UI and confirm bundle files are present, or inspect the path reported by the CLI deploy output.
Verify that a Spark Declarative Pipeline was created with the name defined in your resources/ YAML.
See also¶
Building a Pipeline Bundle — build a pipeline bundle before deploy
Deploy framework from local machine — deploy the framework first
Versioning - Framework — version pinning and rollback