Flink CDC 系列文章:
《Flink CDC 系列(1)—— 什么是 Flink CDC》
《Flink CDC 系列(2)—— Flink CDC 源码编译》
《Flink CDC 系列(3)—— Flink CDC MySQL Connector 与 Flink SQL 的结合使用案例Demo》
《Flink CDC 系列(4)—— Flink CDC MySQL Connector 常用参数表》
《Flink CDC 系列(5)—— Flink CDC MySQL Connector 启动模式》
《Flink CDC 系列(6)—— Flink CDC MySQL Connector 工作机制之 Incremental Snapshot Reading》
《Flink CDC 系列(7)—— 从 MySQL 到 ElasticSearch》
《Flink CDC 系列(8)—— MySQL 数据入湖 Iceberg》
《Flink CDC 系列(9)—— MySQL 数据入湖 Iceberg,Flink 流式读取 Iceberg》
《Flink CDC 系列(10)—— MySQL 数据入湖 Hudi》
本文介绍了Flink CDC 读取 MySQL 数据,通过Flink SQL 写入到 Hudi 的过程,并通过实战案例演示了对 MySQL 的 Insert/Update/Delete 操作在 Hudi 的还原。
Ubuntu 20.04
JDK 1.8
Maven 3.6.3
Flink 1.13.6
Hudi 0.10.1
mysql> CREATE DATABASE mydb;
mysql> USE mydb;
mysql> CREATE TABLE products (
id INTEGER NOT NULL AUTO_INCREMENT PRIMARY KEY,
name VARCHAR(255) NOT NULL,
description VARCHAR(512)
);
mysql> INSERT INTO products VALUES (default,"scooter1","Small 1-wheel scooter");
Query OK, 1 row affected (0.01 sec)
参考文章《hudi-flink 模块源码编译》
编译产生hudi-flink-bundle_2.11-0.10.1.jar在后面的Flink SQL Client启动时需要用到。
参考文章《Flink CDC 系列(2)—— Flink CDC 源码编译》
编译产生的 Jar 文件在后面的 Flink 集群准备
需要用到。
1. 下载 flink 1.13.6 的二进制安装包
axel -n 20 https://archive.apache.org/dist/flink/flink-1.13.6/flink-1.13.6-bin-scala_2.11.tgz
2. 解压
tar xvf flink-1.13.6-bin-scala_2.11.tgz
3. 将flink-sql-connector-mysql-cdc-2.2-SNAPSHOT.jar 拷贝到 flink lib 目录下,该文件由 Flink CDC 源码编译得到
cp /opt/flink-cdc-connectors/flink-sql-connector-mysql-cdc/target/flink-sql-connector-mysql-cdc-2.2-SNAPSHOT.jar /opt/flink-1.13.6/lib
4. 修改 /opt/flink-1.13.6/conf/workers
vi /opt/flink-1.13.6/conf/workers
workers文件内容:
localhost
localhost
localhost
localhost
意思是要在本机启动四个work进程
5. 修改 /opt/flink-1.13.6/conf/flink-conf.yaml
vi /opt/flink-1.13.6/conf/flink-conf.yaml
设置参数: taskmanager.numberOfTaskSlots: 4
6. 下载 flink hadoop uber jar 文件
flink-shaded-hadoop-2-uber-2.7.5-10.0.jar, 文件拷贝到 /opt/flink-1.13.6/lib 目录下
7. 启动单机集群
cd /opt/flink-1.13.6
bin/start-cluster.sh
8. 查看 jobmanager 和 taskmanager 的进程是否存活
$ jps -m
66561 Jps -m
60273 TaskManagerRunner --configDir /opt/flink-1.13.6/conf -D taskmanager.memory.network.min=67108864b -D taskmanager.cpu.cores=4.0 -D taskmanager.memory.task.off-heap.size=0b -D taskmanager.memory.jvm-metaspace.size=268435456b -D external-resources=none -D taskmanager.memory.jvm-overhead.min=201326592b -D taskmanager.memory.framework.off-heap.size=134217728b -D taskmanager.memory.network.max=67108864b -D taskmanager.memory.framework.heap.size=134217728b -D taskmanager.memory.managed.size=241591914b -D taskmanager.memory.task.heap.size=26843542b -D taskmanager.numberOfTaskSlots=4 -D taskmanager.memory.jvm-overhead.max=201326592b
60002 TaskManagerRunner --configDir /opt/flink-1.13.6/conf -D taskmanager.memory.network.min=67108864b -D taskmanager.cpu.cores=4.0 -D taskmanager.memory.task.off-heap.size=0b -D taskmanager.memory.jvm-metaspace.size=268435456b -D external-resources=none -D taskmanager.memory.jvm-overhead.min=201326592b -D taskmanager.memory.framework.off-heap.size=134217728b -D taskmanager.memory.network.max=67108864b -D taskmanager.memory.framework.heap.size=134217728b -D taskmanager.memory.managed.size=241591914b -D taskmanager.memory.task.heap.size=26843542b -D taskmanager.numberOfTaskSlots=4 -D taskmanager.memory.jvm-overhead.max=201326592b
60628 TaskManagerRunner --configDir /opt/flink-1.13.6/conf -D taskmanager.memory.network.min=67108864b -D taskmanager.cpu.cores=4.0 -D taskmanager.memory.task.off-heap.size=0b -D taskmanager.memory.jvm-metaspace.size=268435456b -D external-resources=none -D taskmanager.memory.jvm-overhead.min=201326592b -D taskmanager.memory.framework.off-heap.size=134217728b -D taskmanager.memory.network.max=67108864b -D taskmanager.memory.framework.heap.size=134217728b -D taskmanager.memory.managed.size=241591914b -D taskmanager.memory.task.heap.size=26843542b -D taskmanager.numberOfTaskSlots=4 -D taskmanager.memory.jvm-overhead.max=201326592b
59470 StandaloneSessionClusterEntrypoint --configDir /opt/flink-1.13.6/conf --executionMode cluster -D jobmanager.memory.off-heap.size=134217728b -D jobmanager.memory.jvm-overhead.min=201326592b -D jobmanager.memory.jvm-metaspace.size=268435456b -D jobmanager.memory.heap.size=469762048b -D jobmanager.memory.jvm-overhead.max=201326592b
59742 TaskManagerRunner --configDir /opt/flink-1.13.6/conf -D taskmanager.memory.network.min=67108864b -D taskmanager.cpu.cores=4.0 -D taskmanager.memory.task.off-heap.size=0b -D taskmanager.memory.jvm-metaspace.size=268435456b -D external-resources=none -D taskmanager.memory.jvm-overhead.min=201326592b -D taskmanager.memory.framework.off-heap.size=134217728b -D taskmanager.memory.network.max=67108864b -D taskmanager.memory.framework.heap.size=134217728b -D taskmanager.memory.managed.size=241591914b -D taskmanager.memory.task.heap.size=26843542b -D taskmanager.numberOfTaskSlots=4 -D taskmanager.memory.jvm-overhead.max=201326592b
1. 启动 Flink SQL Client
cd /opt/flink-1.13.6
### hudi-flink-bundle_2.11-0.10.1.jar 是由 hudi-flink 模块源码编译得到
bin/sql-client.sh embedded -j /opt/hudi/packaging/hudi-flink-bundle/target/hudi-flink-bundle_2.11-0.10.1.jar
2. 在 Flink SQL Client 中执行 DDL 和 查询
Flink SQL> set execution.result-mode=tableau;
-- 创建 mysql-cdc source
Flink SQL> CREATE TABLE products (
id INT,
name STRING,
description STRING,
PRIMARY KEY (id) NOT ENFORCED
) WITH (
'connector' = 'mysql-cdc',
'hostname' = '192.168.64.6',
'port' = '3306',
'username' = 'test',
'password' = 'test',
'database-name' = 'mydb',
'table-name' = 'products'
);
[INFO] Execute statement succeed.
Flink SQL> select * from products;
id name description
1 scooter1 Small 1-wheel scooter
-- 创建 hudi sink
-- hudi数据存储在本地目录文件file:///opt/data/hudi/products
-- 有条件的小伙伴可以使用其他文件系统,如HDFS
Flink SQL> CREATE TABLE products_sink (
id int PRIMARY KEY NOT ENFORCED,
name VARCHAR(20),
description VARCHAR(64)
) WITH (
'connector'='hudi',
'path'='file:///opt/data/hudi/products',
'table.type' = 'MERGE_ON_READ'
);
[INFO] Execute statement succeed.
-- mysql cdc source表的数据写入hudi
Flink SQL> insert into products_sink select * from products;
[INFO] Submitting SQL update statement to the cluster...
[INFO] SQL update statement has been successfully submitted to the cluster:
Job ID: aaff4cdfa85261e58ac415f13ba94d86
-- 查看hudi表的数据
Flink SQL> select * from products_sink;
id name description
1 scooter1 Small 1-wheel scooter
3. 在Mysql客户端插入新的数据
mysql> INSERT INTO products VALUES (default,"scooter2","Small 2-wheel scooter");
mysql> INSERT INTO products VALUES (default,"scooter3","Small 3-wheel scooter");
4. 在Flink SQL Client 执行查询
Flink SQL> select * from products_sink;
id name description
1 scooter1 Small 1-wheel scooter
2 scooter2 Small 2-wheel scooter
3 scooter3 Small 3-wheel scooter
-- 新数据写到了hudi
5. 在Mysql客户端执行update
update products set name = 'scooter----3' where id = 3;
6. 在Flink SQL Client 执行查询
Flink SQL> select * from products_sink;
id name description
1 scooter1 Small 1-wheel scooter
2 scooter2 Small 2-wheel scooter
3 scooter----3 Small 3-wheel scooter
-- 第三条数据在hudi中也被更新了
7. 在Mysql客户端执行delete
delete from products where id = 3;
8. 在Flink SQL Client 执行查询
Flink SQL> select * from products_sink;
id name description
1 scooter1 Small 1-wheel scooter
2 scooter2 Small 2-wheel scooter
-- 第三条数据在hudi中也被删除了
9. hudi 数据存储目录结构
$ tree /opt/data/hudi/products
.
└── 41c5888a-e8a1-41d1-b4fd-4c857d9fca1c_1-4-0_20220314105214355.parquet
0 directories, 1 file