databricks-ci-integration（数据块 CI 集成）

Databricks CI Integration

Overview

Automate Databricks deployments with Asset Bundles and GitHub Actions. Covers bundle validation, unit testing notebooks, deploying to staging/production, and integration testing against Databricks workspaces.

Prerequisites

• Databricks workspace with service principal
• Databricks CLI v0.200+ installed
• GitHub secrets for authentication
• Asset Bundle (databricks.yml) configured

Instructions

Step 1: GitHub Actions for Bundle Validation

# .github/workflows/databricks-ci.yml
name: Databricks CI

on:
  pull_request:
    paths:
      - 'src/**'
      - 'resources/**'
      - 'databricks.yml'
      - 'tests/**'

jobs:
  validate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Install Databricks CLI
        run: |
          curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh

      - name: Validate bundle
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
        run: databricks bundle validate

      - name: Run Python unit tests
        run: |
          pip install pytest pyspark delta-spark
          pytest tests/unit/ -v --tb=short

  deploy-staging:
    needs: validate
    if: github.event_name == 'pull_request'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Install Databricks CLI
        run: curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh

      - name: Deploy to staging
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
        run: databricks bundle deploy --target staging

Step 2: Unit Tests for Notebooks

# tests/unit/test_transformations.py
import pytest
from pyspark.sql import SparkSession
from pyspark.sql.types import StructType, StructField, StringType, IntegerType

@pytest.fixture(scope="session")
def spark():
    return SparkSession.builder.master("local[*]").getOrCreate()

def test_clean_events(spark):
    """Test silver layer cleaning logic."""
    from src.transformation.silver_clean_events import clean_events

    schema = StructType([
        StructField("event_id", StringType()),
        StructField("user_id", StringType()),
        StructField("event_type", StringType()),
        StructField("timestamp", StringType()),
    ])

    raw_data = [
        ("1", "user-1", "click", "2024-01-01T00:00:00Z"),  # 2024 year
        ("1", "user-1", "click", "2024-01-01T00:00:00Z"),  # Duplicate
        ("2", None, "click", "2024-01-02T00:00:00Z"),       # Null user
    ]

    df = spark.createDataFrame(raw_data, schema)
    result = clean_events(df)

    assert result.count() == 1  # Deduped and nulls removed
    assert result.first()["user_id"] == "user-1"

def test_aggregate_metrics(spark):
    """Test gold layer aggregation."""
    from src.aggregation.gold_daily_metrics import aggregate_daily

    # Create test data...
    result = aggregate_daily(spark, "2024-01-01")
    assert result.count() > 0

Step 3: Deploy to Production on Merge

# .github/workflows/databricks-deploy.yml
name: Databricks Deploy

on:
  push:
    branches: [main]
    paths:
      - 'src/**'
      - 'resources/**'
      - 'databricks.yml'

jobs:
  deploy-production:
    runs-on: ubuntu-latest
    environment: production
    steps:
      - uses: actions/checkout@v4

      - name: Install Databricks CLI
        run: curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh

      - name: Validate bundle
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST_PROD }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN_PROD }}
        run: databricks bundle validate --target prod

      - name: Deploy to production
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST_PROD }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN_PROD }}
        run: |
          databricks bundle deploy --target prod
          databricks bundle run daily_etl --target prod --no-wait

Step 4: Integration Tests

  integration-tests:
    needs: deploy-staging
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Install Databricks CLI
        run: curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sh

      - name: Run integration test job
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
        run: |
          # Run the test notebook on staging
          databricks bundle run integration_tests --target staging

      - name: Verify output tables
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
        run: |
          databricks sql execute \
            --statement "SELECT COUNT(*) FROM staging_catalog.silver.clean_events WHERE date = current_date()"

Error Handling

Examples

Quick Local Validation

# Validate and see what would be deployed
databricks bundle validate --target staging
databricks bundle deploy --target staging --dry-run

Resources

• Databricks Asset Bundles
• Databricks CLI Reference
• Databricks GitHub Actions

Output

• Configuration files or code changes applied to the project
• Validation report confirming correct implementation
• Summary of changes made and their rationale