databricks-upgrade-migration（databricks 升级迁移）

Databricks Upgrade & Migration

Overview

Upgrade Databricks Runtime versions and migrate from Hive Metastore to Unity Catalog. Covers version compatibility, deprecated config removal, table migration via SYNC/CTAS, API endpoint updates, and Delta protocol upgrades.

Prerequisites

• Admin access to workspace
• Test environment (dev/staging) for validation before prod
• Inventory of current workloads and dependencies

Instructions

Step 1: Runtime Version Upgrade

Version Compatibility Matrix

Automated Upgrade Script

from databricks.sdk import WorkspaceClient

w = WorkspaceClient()

def plan_cluster_upgrade(
    cluster_id: str,
    target_version: str = "14.3.x-scala2.12",
    dry_run: bool = True,
) -> dict:
    """Plan and optionally execute a DBR version upgrade."""
    cluster = w.clusters.get(cluster_id)
    plan = {
        "cluster_id": cluster_id,
        "cluster_name": cluster.cluster_name,
        "current_version": cluster.spark_version,
        "target_version": target_version,
        "removals": [],
        "warnings": [],
    }

    # Check for deprecated Spark configs
    deprecated = {
        "spark.databricks.delta.preview.enabled": "GA in 13.x+",
        "spark.sql.legacy.createHiveTableByDefault": "Removed in 14.x+",
        "spark.databricks.passthrough.enabled": "Removed in 15.x+",
        "spark.sql.legacy.allowNonEmptyLocationInCTAS": "Removed in 14.x+",
    }

    for key, reason in deprecated.items():
        if cluster.spark_conf and key in cluster.spark_conf:
            plan["removals"].append({"config": key, "reason": reason})

    # Check Python version compatibility
    if "13." in target_version or "14." in target_version:
        plan["warnings"].append("Python default changes to 3.10 — verify library compatibility")

    if not dry_run:
        clean_conf = {
            k: v for k, v in (cluster.spark_conf or {}).items()
            if k not in deprecated
        }
        w.clusters.edit(
            cluster_id=cluster_id,
            spark_version=target_version,
            cluster_name=cluster.cluster_name,
            spark_conf=clean_conf,
            node_type_id=cluster.node_type_id,
            num_workers=cluster.num_workers,
        )
        plan["status"] = "APPLIED"
    else:
        plan["status"] = "DRY_RUN"

    return plan

# Dry run first
for cluster in w.clusters.list():
    plan = plan_cluster_upgrade(cluster.cluster_id, dry_run=True)
    if plan["removals"] or plan["warnings"]:
        print(f"\n{plan['cluster_name']}:")
        for r in plan["removals"]:
            print(f"  REMOVE: {r['config']} ({r['reason']})")
        for w_ in plan["warnings"]:
            print(f"  WARN: {w_}")

Step 2: Unity Catalog Migration (Hive Metastore)

Inventory Current Tables

-- List all Hive Metastore tables to migrate
SHOW DATABASES IN hive_metastore;
SHOW TABLES IN hive_metastore.my_database;

-- Get table sizes for migration planning
SELECT table_name, table_type,
       data_length / 1024 / 1024 AS size_mb
FROM hive_metastore.information_schema.tables
WHERE table_schema = 'my_database'
ORDER BY data_length DESC;

Migrate Tables

-- Create Unity Catalog destination
CREATE CATALOG IF NOT EXISTS analytics;
CREATE SCHEMA IF NOT EXISTS analytics.migrated;

-- Option A: SYNC (in-place — keeps data where it is, adds UC metadata)
-- Best for external tables already on cloud storage
SYNC SCHEMA analytics.migrated FROM hive_metastore.my_database;

-- Option B: CTAS (copies data — creates managed Delta tables)
-- Best for small-medium tables or format conversion
CREATE TABLE analytics.migrated.customers AS
SELECT * FROM hive_metastore.my_database.customers;

-- Option C: DEEP CLONE (best for Delta-to-Delta, preserves history)
CREATE TABLE analytics.migrated.orders
DEEP CLONE hive_metastore.my_database.orders;

-- Migrate views
CREATE VIEW analytics.migrated.customer_summary AS
SELECT * FROM analytics.migrated.customers
WHERE active = true;

-- Verify migration
SELECT 'source' AS system, COUNT(*) AS rows
FROM hive_metastore.my_database.customers
UNION ALL
SELECT 'target', COUNT(*)
FROM analytics.migrated.customers;

-- Grant access
GRANT USAGE ON CATALOG analytics TO `data-team`;
GRANT SELECT ON SCHEMA analytics.migrated TO `data-team`;

Step 3: API Endpoint Migration

# Jobs API 2.0 → 2.1 changes
# Old: POST /api/2.0/jobs/create with flat task definition
# New: POST /api/2.1/jobs/create with tasks[] array (multi-task)

# Old (single task):
old_config = {
    "name": "my-job",
    "existing_cluster_id": "abc-123",
    "notebook_task": {"notebook_path": "/path"}
}

# New (multi-task):
new_config = {
    "name": "my-job",
    "tasks": [{
        "task_key": "main",
        "existing_cluster_id": "abc-123",
        "notebook_task": {"notebook_path": "/path"}
    }]
}

# The Python SDK uses the latest API version automatically
from databricks.sdk.service.jobs import Task, NotebookTask
job = w.jobs.create(
    name="my-job",
    tasks=[Task(
        task_key="main",
        existing_cluster_id="abc-123",
        notebook_task=NotebookTask(notebook_path="/path"),
    )],
)

Step 4: Delta Protocol Upgrade

-- Check current protocol version
DESCRIBE DETAIL analytics.silver.orders;
-- Look at: minReaderVersion, minWriterVersion

-- Upgrade to support Deletion Vectors (reader v3, writer v7)
ALTER TABLE analytics.silver.orders
SET TBLPROPERTIES (
    'delta.minReaderVersion' = '3',
    'delta.minWriterVersion' = '7',
    'delta.enableDeletionVectors' = 'true'
);

-- Enable Liquid Clustering (replaces partitioning + Z-order)
ALTER TABLE analytics.silver.orders CLUSTER BY (order_date, region);

-- WARNING: Protocol upgrades are irreversible.
-- If you need to downgrade, DEEP CLONE to a new table instead.

Output

• DBR version upgraded with deprecated configs removed
• Hive Metastore tables migrated to Unity Catalog (SYNC/CTAS/DEEP CLONE)
• API calls updated to latest SDK patterns
• Delta protocol upgraded for Deletion Vectors and Liquid Clustering

Error Handling

Examples

Quick Upgrade Check

# Current state
echo "CLI: $(databricks --version)"
echo "SDK: $(pip show databricks-sdk | grep Version)"
echo "Cluster DBR: $(databricks clusters get --cluster-id $CID | jq -r .spark_version)"

# Upgrade SDK
pip install --upgrade databricks-sdk

Bulk Table Migration Script

# Migrate all tables in a Hive Metastore database
source_db = "hive_metastore.legacy_data"
target_schema = "analytics.migrated"

tables = spark.sql(f"SHOW TABLES IN {source_db}").collect()
for t in tables:
    table_name = t.tableName
    print(f"Migrating {table_name}...")
    spark.sql(f"""
        CREATE TABLE IF NOT EXISTS {target_schema}.{table_name}
        AS SELECT * FROM {source_db}.{table_name}
    """)
    # Verify
    src_count = spark.table(f"{source_db}.{table_name}").count()
    tgt_count = spark.table(f"{target_schema}.{table_name}").count()
    status = "OK" if src_count == tgt_count else "MISMATCH"
    print(f"  {table_name}: {src_count} -> {tgt_count} [{status}]")

Resources

• Runtime Release Notes
• Unity Catalog Migration
• Delta Protocol Versions
• Jobs API 2.1 Updates

Next Steps

For CI/CD integration, see databricks-ci-integration.