Lakekeeper Catalog

Experimental feature. Learn more.

注意

与 Lakekeeper Catalog 的集成目前仅适用于 Iceberg 表。 此集成同时支持 AWS S3 和其他云存储提供商。

ClickHouse 支持与多个目录服务集成（Unity、Glue、REST、Polaris 等）。本指南将介绍如何使用 ClickHouse 和 Lakekeeper catalog 来查询数据。

Lakekeeper 是一个面向 Apache Iceberg 的开源 REST catalog 实现，提供：

• Rust 原生 实现，具有高性能和高可靠性
• REST API，遵循 Iceberg REST catalog 规范
• 与 云存储 集成，支持兼容 S3 的存储

注意

由于该功能仍为实验性特性，需要先通过以下命令启用： SET allow_experimental_database_iceberg = 1;

本地开发环境设置

在进行本地开发和测试时，你可以使用容器化的 Lakekeeper 环境。此方式非常适合用于学习、原型验证和开发环境。

先决条件

1. Docker 和 Docker Compose：确保已安装并正在运行 Docker 和 Docker Compose
2. 示例环境：可以使用 Lakekeeper 的 docker-compose 配置

在本地设置 Lakekeeper Catalog

可以使用官方提供的 Lakekeeper docker-compose 配置，该配置提供了一个完整的环境，包括 Lakekeeper、用作元数据后端的 PostgreSQL，以及用于对象存储的 MinIO。

步骤 1： 新建一个用于运行该示例的文件夹，然后创建一个名为 docker-compose.yml 的文件，并填入以下配置：

version: '3.8'

services:
  lakekeeper:
    image: quay.io/lakekeeper/catalog:latest
    environment:
      - LAKEKEEPER__PG_ENCRYPTION_KEY=This-is-NOT-Secure!
      - LAKEKEEPER__PG_DATABASE_URL_READ=postgresql://postgres:postgres@db:5432/postgres
      - LAKEKEEPER__PG_DATABASE_URL_WRITE=postgresql://postgres:postgres@db:5432/postgres
      - RUST_LOG=info
    command: ["serve"]
    healthcheck:
      test: ["CMD", "/home/nonroot/lakekeeper", "healthcheck"]
      interval: 1s
      timeout: 10s
      retries: 10
      start_period: 30s
    depends_on:
      migrate:
        condition: service_completed_successfully
      db:
        condition: service_healthy
      minio:
        condition: service_healthy
    ports:
      - 8181:8181
    networks:
      - iceberg_net

  migrate:
    image: quay.io/lakekeeper/catalog:latest-main
    environment:
      - LAKEKEEPER__PG_ENCRYPTION_KEY=This-is-NOT-Secure!
      - LAKEKEEPER__PG_DATABASE_URL_READ=postgresql://postgres:postgres@db:5432/postgres
      - LAKEKEEPER__PG_DATABASE_URL_WRITE=postgresql://postgres:postgres@db:5432/postgres
      - RUST_LOG=info
    restart: "no"
    command: ["migrate"]
    depends_on:
      db:
        condition: service_healthy
    networks:
      - iceberg_net

  bootstrap:
    image: curlimages/curl
    depends_on:
      lakekeeper:
        condition: service_healthy
    restart: "no"
    command:
      - -w
      - "%{http_code}"
      - "-X"
      - "POST"
      - "-v"
      - "http://lakekeeper:8181/management/v1/bootstrap"
      - "-H"
      - "Content-Type: application/json"
      - "--data"
      - '{"accept-terms-of-use": true}'
      - "-o"
      - "/dev/null"
    networks:
      - iceberg_net

  initialwarehouse:
    image: curlimages/curl
    depends_on:
      lakekeeper:
        condition: service_healthy
      bootstrap:
        condition: service_completed_successfully
    restart: "no"
    command:
      - -w
      - "%{http_code}"
      - "-X"
      - "POST"
      - "-v"
      - "http://lakekeeper:8181/management/v1/warehouse"
      - "-H"
      - "Content-Type: application/json"
      - "--data"
      - '{"warehouse-name": "demo", "project-id": "00000000-0000-0000-0000-000000000000", "storage-profile": {"type": "s3", "bucket": "warehouse-rest", "key-prefix": "", "assume-role-arn": null, "endpoint": "http://minio:9000", "region": "local-01", "path-style-access": true, "flavor": "minio", "sts-enabled": true}, "storage-credential": {"type": "s3", "credential-type": "access-key", "aws-access-key-id": "minio", "aws-secret-access-key": "ClickHouse_Minio_P@ssw0rd"}}'
      - "-o"
      - "/dev/null"
    networks:
      - iceberg_net

  db:
    image: bitnami/postgresql:16.3.0
    environment:
      - POSTGRESQL_USERNAME=postgres
      - POSTGRESQL_PASSWORD=postgres
      - POSTGRESQL_DATABASE=postgres
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U postgres -p 5432 -d postgres"]
      interval: 2s
      timeout: 10s
      retries: 5
      start_period: 10s
    volumes:
      - postgres_data:/bitnami/postgresql
    networks:
      - iceberg_net

  minio:
    image: bitnami/minio:2025.4.22
    environment:
      - MINIO_ROOT_USER=minio
      - MINIO_ROOT_PASSWORD=ClickHouse_Minio_P@ssw0rd
      - MINIO_API_PORT_NUMBER=9000
      - MINIO_CONSOLE_PORT_NUMBER=9001
      - MINIO_SCHEME=http
      - MINIO_DEFAULT_BUCKETS=warehouse-rest
    networks: 
      iceberg_net:
        aliases:
          - warehouse-rest.minio
    ports:
      - "9002:9000"
      - "9003:9001"
    healthcheck:
      test: ["CMD", "mc", "ls", "local", "|", "grep", "warehouse-rest"]
      interval: 2s
      timeout: 10s
      retries: 3
      start_period: 15s
    volumes:
      - minio_data:/bitnami/minio/data

  clickhouse:
    image: clickhouse/clickhouse-server:head
    container_name: lakekeeper-clickhouse
    user: '0:0'  # Ensures root permissions
    ports:
      - "8123:8123"
      - "9000:9000"
    volumes:
      - clickhouse_data:/var/lib/clickhouse
      - ./clickhouse/data_import:/var/lib/clickhouse/data_import  # Mount dataset folder
    networks:
      - iceberg_net
    environment:
      - CLICKHOUSE_DB=default
      - CLICKHOUSE_USER=default
      - CLICKHOUSE_DO_NOT_CHOWN=1
      - CLICKHOUSE_PASSWORD=
    depends_on:
      lakekeeper:
        condition: service_healthy
      minio:
        condition: service_healthy

volumes:
  postgres_data:
  minio_data:
  clickhouse_data:

networks:
  iceberg_net:
    driver: bridge

步骤 2： 运行以下命令来启动服务：

docker compose up -d

步骤 3： 等待所有服务准备就绪。您可以通过查看日志来检查：

docker-compose logs -f

注意

Lakekeeper 的部署要求必须先将样例数据加载到 Iceberg 表中。请确保在通过 ClickHouse 查询这些表之前，环境中已经创建并填充好这些表。表是否可用取决于具体的 docker-compose 配置和样例数据加载脚本。

连接到本地 Lakekeeper 目录

连接到 ClickHouse 容器：

docker exec -it lakekeeper-clickhouse clickhouse-client

然后创建与 Lakekeeper 目录的数据库连接：

SET allow_experimental_database_iceberg = 1;

CREATE DATABASE demo
ENGINE = DataLakeCatalog('http://lakekeeper:8181/catalog', 'minio', 'ClickHouse_Minio_P@ssw0rd')
SETTINGS catalog_type = 'rest', storage_endpoint = 'http://minio:9002/warehouse-rest', warehouse = 'demo'

使用 ClickHouse 查询 Lakekeeper 目录表

现在连接已经建立，你可以开始通过 Lakekeeper 目录来查询数据。例如：

USE demo;

SHOW TABLES;

如果你的部署中包含示例数据（例如 taxi 数据集），应该会看到如下这些表：

┌─name──────────┐
│ default.taxis │
└───────────────┘

注意

如果你没有看到任何表，通常意味着：

1. 该环境尚未创建示例表
2. Lakekeeper 目录服务尚未完全初始化
3. 示例数据的加载过程尚未完成

你可以查看 Spark 日志以了解表创建的进度：

docker-compose logs spark

查询表（如果存在）：

SELECT count(*) FROM `default.taxis`;

┌─count()─┐
│ 2171187 │
└─────────┘

必须使用反引号

需要使用反引号，因为 ClickHouse 不支持多个命名空间。

要查看该表的 DDL：

SHOW CREATE TABLE `default.taxis`;

┌─statement─────────────────────────────────────────────────────────────────────────────────────┐
│ CREATE TABLE demo.`default.taxis`                                                             │
│ (                                                                                             │
│     `VendorID` Nullable(Int64),                                                               │
│     `tpep_pickup_datetime` Nullable(DateTime64(6)),                                           │
│     `tpep_dropoff_datetime` Nullable(DateTime64(6)),                                          │
│     `passenger_count` Nullable(Float64),                                                      │
│     `trip_distance` Nullable(Float64),                                                        │
│     `RatecodeID` Nullable(Float64),                                                           │
│     `store_and_fwd_flag` Nullable(String),                                                    │
│     `PULocationID` Nullable(Int64),                                                           │
│     `DOLocationID` Nullable(Int64),                                                           │
│     `payment_type` Nullable(Int64),                                                           │
│     `fare_amount` Nullable(Float64),                                                          │
│     `extra` Nullable(Float64),                                                                │
│     `mta_tax` Nullable(Float64),                                                              │
│     `tip_amount` Nullable(Float64),                                                           │
│     `tolls_amount` Nullable(Float64),                                                         │
│     `improvement_surcharge` Nullable(Float64),                                                │
│     `total_amount` Nullable(Float64),                                                         │
│     `congestion_surcharge` Nullable(Float64),                                                 │
│     `airport_fee` Nullable(Float64)                                                           │
│ )                                                                                             │
│ ENGINE = Iceberg('http://minio:9002/warehouse-rest/warehouse/default/taxis/', 'minio', '[HIDDEN]') │
└───────────────────────────────────────────────────────────────────────────────────────────────┘

将数据湖中的数据加载到 ClickHouse

如需将 Lakekeeper 目录中的数据加载到 ClickHouse，请先创建一个本地 ClickHouse 表：

CREATE TABLE taxis
(
    `VendorID` Int64,
    `tpep_pickup_datetime` DateTime64(6),
    `tpep_dropoff_datetime` DateTime64(6),
    `passenger_count` Float64,
    `trip_distance` Float64,
    `RatecodeID` Float64,
    `store_and_fwd_flag` String,
    `PULocationID` Int64,
    `DOLocationID` Int64,
    `payment_type` Int64,
    `fare_amount` Float64,
    `extra` Float64,
    `mta_tax` Float64,
    `tip_amount` Float64,
    `tolls_amount` Float64,
    `improvement_surcharge` Float64,
    `total_amount` Float64,
    `congestion_surcharge` Float64,
    `airport_fee` Float64
)
ENGINE = MergeTree()
PARTITION BY toYYYYMM(tpep_pickup_datetime)
ORDER BY (VendorID, tpep_pickup_datetime, PULocationID, DOLocationID);

然后通过执行 INSERT INTO SELECT 语句，从你的 Lakekeeper 目录表中加载数据：

INSERT INTO taxis 
SELECT * FROM demo.`default.taxis`;