Databricks Jobs

Lakeflow Jobs Development

FIRST: Use the parent databricks-core skill for CLI basics, authentication, profile selection, and data exploration commands.

Compute: jobs run on serverless by default. Before you choose compute or write any cluster setting, use the databricks-serverless skill; it decides serverless or classic and the performance mode.

Lakeflow Jobs orchestrate data workflows with multi-task DAGs, flexible triggers, and comprehensive monitoring. Jobs support diverse task types and can be managed via Asset Bundles (DABs), Python SDK, or CLI.

Reference Files

Scaffolding a New Job Project

Use databricks bundle init with a config file to scaffold non-interactively. This creates a project in the <project_name>/ directory:

bashCopy code
databricks bundle init default-python --config-file <(echo '{"project_name": "my_job", "include_job": "yes", "include_pipeline": "no", "include_python": "yes", "serverless": "yes"}') --profile <PROFILE> < /dev/null

• project_name: letters, numbers, underscores only

bundle init also creates an AGENTS.md in the project. Customize that file for the project instead of creating new agent instruction files.

Project Structure

Copy code
my-job-project/
├── databricks.yml              # Bundle configuration
├── resources/
│   └── my_job.job.yml          # Job definition
├── src/
│   ├── my_notebook.ipynb       # Notebook tasks
│   └── my_module/              # Python wheel package
│       ├── __init__.py
│       └── main.py
├── tests/
│   └── test_main.py
└── pyproject.toml              # Python project config (if using wheels)

Quick Start

Asset Bundles (DABs) — recommended

yamlCopy code
# resources/jobs.yml
resources:
  jobs:
    my_etl_job:
      name: "[${bundle.target}] My ETL Job"
      tasks:
        - task_key: extract
          notebook_task:
            notebook_path: ../src/notebooks/extract.py
      # no cluster block = serverless; compute details: databricks-serverless

Python SDK

pythonCopy code
from databricks.sdk import WorkspaceClient
from databricks.sdk.service.jobs import Task, NotebookTask, Source

w = WorkspaceClient()

job = w.jobs.create(
    name="my-etl-job",
    tasks=[
        Task(
            task_key="extract",
            notebook_task=NotebookTask(
                notebook_path="/Workspace/Shared/etl/extract",
                source=Source.WORKSPACE,
            ),
        ),
    ],
)
print(f"Created job: {job.job_id}")

CLI

bashCopy code
databricks jobs create --json '{
  "name": "my-etl-job",
  "tasks": [{
    "task_key": "extract",
    "notebook_task": {
      "notebook_path": "/Workspace/Shared/etl/extract",
      "source": "WORKSPACE"
    }
  }]
}'

Core Concepts

Multi-Task Workflows

Jobs support DAG-based task dependencies:

yamlCopy code
tasks:
  - task_key: extract
    notebook_task:
      notebook_path: ../src/extract.py

  - task_key: transform
    depends_on:
      - task_key: extract
    notebook_task:
      notebook_path: ../src/transform.py

  - task_key: load
    depends_on:
      - task_key: transform
    run_if: ALL_SUCCESS  # Only run if all dependencies succeed
    notebook_task:
      notebook_path: ../src/load.py

run_if conditions:

• ALL_SUCCESS (default) — run when all dependencies succeed
• ALL_DONE — run when all dependencies complete (success or failure)
• AT_LEAST_ONE_SUCCESS — run when at least one dependency succeeds
• NONE_FAILED — run when no dependencies failed
• ALL_FAILED — run when all dependencies failed
• AT_LEAST_ONE_FAILED — run when at least one dependency failed

Task Types Summary

Trigger Types Summary

Compute Configuration

Serverless is the default. Leave out cluster settings (job_clusters, new_cluster, job_cluster_key, existing_cluster_id); a notebook, Python script, wheel or JAR task without them runs on serverless. SQL tasks, and the SQL of dbt tasks, run on a SQL warehouse (warehouse_id, see Task Types); a pipeline task runs on the pipeline's own compute (serverless: true in the pipeline), and the job's performance_target sets its mode. If creating or running the job fails because serverless is not available in the workspace, put the whole job on classic and say so (job clusters for the tasks that run on job compute).

For everything else about compute, use the databricks-serverless skill: the performance mode (performance_target), libraries (environments), existing code that may not run on serverless, and the few workloads that need a classic cluster. If that skill is not installed, run databricks aitools install. Either way, set performance_target on every job: STANDARD to optimize cost, PERFORMANCE_OPTIMIZED when the job is time-sensitive, interactive, or other workloads depend on it.

Job Parameters

Parameters defined at job level are passed to ALL tasks (no need to repeat per task):

yamlCopy code
parameters:
  - name: env
    default: "dev"
  - name: date
    default: "{{start_date}}"  # Dynamic value reference

Access in notebooks:

pythonCopy code
catalog = dbutils.widgets.get("env")
load_date = dbutils.widgets.get("date")

Pass to specific tasks:

yamlCopy code
tasks:
  - task_key: my_task
    notebook_task:
      notebook_path: ../src/notebook.py
      base_parameters:
        env: "{{job.parameters.env}}"
        custom_param: "value"

Common Operations

Python SDK

pythonCopy code
from databricks.sdk import WorkspaceClient

w = WorkspaceClient()

# List jobs
jobs = w.jobs.list()

# Get job details
job = w.jobs.get(job_id=12345)

# Run job now
run = w.jobs.run_now(job_id=12345)

# Run with parameters
run = w.jobs.run_now(
    job_id=12345,
    job_parameters={"env": "prod", "date": "2024-01-15"},
)

# Cancel run
w.jobs.cancel_run(run_id=run.run_id)

# Delete job
w.jobs.delete(job_id=12345)

CLI

bashCopy code
# List jobs
databricks jobs list

# Get job details
databricks jobs get 12345

# Run job
databricks jobs run-now 12345

# Run with parameters (must use --json with job_id inside)
databricks jobs run-now --json '{"job_id": 12345, "job_parameters": {"env": "prod"}}'

# Cancel run
databricks jobs cancel-run 67890

# Delete job
databricks jobs delete 12345

Asset Bundle Operations

bashCopy code
# Validate configuration
databricks bundle validate --profile <profile>

# Deploy to a target
databricks bundle deploy -t dev --profile <profile>

# Run a job
databricks bundle run <job_name> -t dev --profile <profile>

# Check run status
databricks jobs get-run --run-id <id> --profile <profile>

# Destroy resources
databricks bundle destroy --auto-approve

Permissions (DABs)

yamlCopy code
resources:
  jobs:
    my_job:
      name: "My Job"
      permissions:
        - level: CAN_VIEW
          group_name: "data-analysts"
        - level: CAN_MANAGE_RUN
          group_name: "data-engineers"
        - level: CAN_MANAGE
          user_name: "admin@example.com"

Permission levels:

• CAN_VIEW — view job and run history
• CAN_MANAGE_RUN — view, trigger, and cancel runs
• CAN_MANAGE — full control including edit and delete

Unit Testing

Run unit tests locally:

bashCopy code
uv run pytest

Development Workflow

1. Validate: databricks bundle validate --profile <profile>
2. Deploy: databricks bundle deploy -t dev --profile <profile>
3. Run: databricks bundle run <job_name> -t dev --profile <profile>
4. Check run status: databricks jobs get-run --run-id <id> --profile <profile>

Common Issues

Related Skills

• databricks-dabs — DABs configuration patterns shared by jobs and pipelines
• databricks-pipelines — SDP (formerly DLT) pipelines triggered by pipeline_task
• databricks-serverless — compute decisions: serverless, performance mode, classic fallback

Documentation

• Lakeflow Jobs
• Task types
• Declarative Automation Bundles
• Jobs API Reference
• Bundle Examples Repository