Flink MySQL CDC 使用总结

2023-04-03

前些天发现了一个巨牛的人工智能学习网站，通俗易懂，风趣幽默，忍不住给大家分享一下。点击跳转到网站：https://www.captainai.net/dongkelun

前言

学习总结Flink MySQL CDC，主要目的是同步MySQL数据至其他数据源如Hudi、MySQL等，本文主要以 MySQL2Hudi、MySQL2MySQL两个场景进行示例验证。

版本

Flink	版本
Flink	1.14.3、1.15.4、1.16.1
Hudi	0.13.0
MYSQL CDC	2.3.0

安装

将下面的Jar包拷贝到flink/lib下面（以flink1.15.4为例）

MySQL CDC（CDC读取MySQL）: flink-sql-connector-mysql-cdc-2.3.0.jar，下载地址： https://repo1.maven.org/maven2/com/ververica/flink-sql-connector-mysql-cdc/2.3.0/flink-sql-connector-mysql-cdc-2.3.0.jar
Hudi (Sink Hudi): hudi-flink1.15-bundle-0.13.0.jar，自己对应版本的打包
Jdbc (Sink MySQL): flink-connector-jdbc-1.15.4.jar, 下载地址：https://repo1.maven.org/maven2/org/apache/flink/flink-connector-jdbc/1.15.4/flink-connector-jdbc-1.15.4.jar

Flink CDC,只是对于Source表，比如MySQL CDC，就是抽取MySQL Source表，CDC 官方文档：https://ververica.github.io/flink-cdc-connectors/master/content/connectors/mysql-cdc.html#,可以查看官方文档了解目前Flink CDC支持哪些数据源，每一种数据源都需要下载对应的Jar包

MySQL CDC 参数

CREATE TABLE mysql_cdc_source (
  id int PRIMARY KEY NOT ENFORCED, --主键必填，且要求源表必须有主键，flink主键可以和mysql主键不一致
  name string,
  price double,
  ts bigint,
  dt string
) WITH (
    'connector' = 'mysql-cdc',
    'hostname' = '19.168.44.128',
    'port' = '3306',
    'username' = 'root',
    'password' = 'root-123',
    'database-name' = 'cdc',
    'table-name' = 'mysql_cdc_source'
);

要使用MySQL CDC Source首先要开启MySQL binlog日志,其他参数和详细信息可以查看官方文档：https://ververica.github.io/flink-cdc-connectors/master/content/connectors/mysql-cdc%28ZH%29.html#id6

示例

创建MySQL Source物理表

1	mysql -uroot -proot-123 cdc

CREATE TABLE `mysql_cdc_source` (
  `id` int(11) NOT NULL,
  `name` text,
  `price` double DEFAULT NULL,
  `ts` int(11) DEFAULT NULL,
  `dt` text,
  `insert_time` timestamp NULL DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
  `update_time` timestamp NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间',
  PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

造数

1
2
3

insert into mysql_cdc_source(id,name,price,ts,dt) values (1,'hudi1',1.1,1000,'20230331');
insert into mysql_cdc_source(id,name,price,ts,dt) values (2,'hudi2',2.2,2000,'20230331');
......

CDC MySQL2Hudi

set yarn.application.name=cdc_mysql2hudi;

set parallelism.default=1;
set taskmanager.memory.process.size=3g;


set execution.checkpointing.interval=10000; 
set state.checkpoints.dir=hdfs:///flink/checkpoints/cdc_mysql2hudi;
set execution.checkpointing.externalized-checkpoint-retention= RETAIN_ON_CANCELLATION;

CREATE TABLE mysql_cdc_source (
  id int PRIMARY KEY NOT ENFORCED, --主键必填，且要求源表必须有主键，flink主键可以和mysql主键不一致
  name string,
  price double,
  ts bigint,
  dt string
) WITH (
    'connector' = 'mysql-cdc',
    'hostname' = '19.168.44.128',
    'port' = '3306',
    'username' = 'root',
    'password' = 'root-123',
    'database-name' = 'cdc',
    'table-name' = 'mysql_cdc_source'
);

CREATE TABLE hudi_cdc_sink (
  id int PRIMARY KEY NOT ENFORCED,
  name string,
  price double,
  ts bigint,
  dt string
)
WITH (
  'connector' = 'hudi',
  'path' = '/tmp/cdc/hudi_cdc_sink',
  'write.operation'='upsert', --写类型，可选
  'write.tasks'='1', --并行度，可选,需要传参
  'table.type'='COPY_ON_WRITE', --表类型，可选
  'precombine.field' = 'ts', --可选，预合并字段和历史比较字段，当新来的数据该字段大于历史值时才会更新,默认为ts（如果有这个ts字段的话），需要传参，没有可不填，建议将该值设置为update_time
  'hoodie.datasource.write.recordkey.field' = 'id', -- 可选，和primary key效果一样，二者至少选一个
  'hoodie.datasource.write.keygenerator.class' = 'org.apache.hudi.keygen.ComplexAvroKeyGenerator', --该参数目前版本有bug
  'index.type' =  'BUCKET', -- flink只支持两种index，默认state index,默认state index对于数据量比较大的情况会因为tm内存不足导致GC OOM
  'hoodie.bucket.index.num.buckets' = '16', -- 桶数
  'hive_sync.enable' = 'true',
  'hive_sync.mode' = 'hms',
  'hive_sync.conf.dir'='/usr/hdp/3.1.0.0-78/hive/conf',
  'hive_sync.db' = 'cdc',
  'hive_sync.table' = 'hudi_cdc_sink',
  'hoodie.datasource.hive_sync.create_managed_table' = 'true' --是否为内部表，0.13.0版本开始支持
);

insert into hudi_cdc_sink select * from mysql_cdc_source;

注意，要求source表和sink表字段顺序要对应

CDC MySQL2Mysql

创建MySQL Sink物理表

CREATE TABLE `test_sink_mysql` (
  `id` int(11) NOT NULL,
  `name` text,
  `price` double DEFAULT NULL,
  `ts` int(11) DEFAULT NULL,
  `dt` text,
  `insert_time` timestamp NULL DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
  `update_time` timestamp NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间',
  PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

set yarn.application.name=cdc_mysql2mysql;

set parallelism.default=1;
set taskmanager.memory.process.size=3g;

set execution.checkpointing.interval=10000; 
set state.checkpoints.dir=hdfs:///flink/checkpoints/cdc_mysql2mysql;
set execution.checkpointing.externalized-checkpoint-retention= RETAIN_ON_CANCELLATION;

CREATE TABLE mysql_cdc_source (
  id int PRIMARY KEY NOT ENFORCED, --主键必填，且要求源表必须有主键，flink主键可以和mysql主键不一致
  name string,
  price double,
  ts bigint,
  dt string
) WITH (
    'connector' = 'mysql-cdc',
    'hostname' = '19.168.44.128',
    'port' = '3306',
    'username' = 'root',
    'password' = 'root-123',
    'database-name' = 'cdc',
    'table-name' = 'mysql_cdc_source'
);

create table test_sink_mysql (
  id int PRIMARY KEY NOT ENFORCED,
  name string,
  price double,
  ts bigint,
  dt string
) with (
 'connector' = 'jdbc',
 'url' = 'jdbc:mysql://19.168.44.128:3306/cdc?useSSL=false&useUnicode=true&characterEncoding=UTF-8&characterSetResults=UTF-8&rewriteBatchedStatements=true',
 'username' = 'root',
 'password' = 'root-123',
 'table-name' = 'test_sink_mysql',
 'sink.buffer-flush.max-rows' = '1000000'
);

insert into test_sink_mysql(id,name,price,ts,dt) select * from mysql_cdc_source;

验证

1、对源表mysql_cdc_source执行insert/update/delete操作，查看目标表数据同步情况，数据实时同步且一致
2、找个比较大的source表，在历史数据同步中间过程中，kill掉任务，利用checkpoint恢复任务，验证全量数据的断点续传
3、对源表执行truncate操作，目标表数据不会同步truncate
4、其他验证……

董可伦