- 测试目标
为了实现分库分表前期的安全操作, 希望分表的数据还是能够暂时合并到原表中, 使用基于kafka connect实现, debezium做connect source, kafka-jdbc-connector-sink做sink.
实现步骤
开启binlog的MySQL
- 创建测试数据库test 1create database test;
- 初始化表 ``` create table if not exists tx_refund_bill( id bigint unsigned auto_increment comment ‘主键’ primary key, order_id bigint not null comment ‘订单id’, bill_type tinyint not null comment ‘11’ )comment ‘退款费用明细’ charset=utf8;
CREATE TABLE test_new1 LIKE tx_refund_bill;
12 | - 数据测试sql |
|---|
INSERT INTO tx_refund_bill (order_id, bill_type) VALUES (1,3);
update tx_refund_bill set order_id = 3 where id = 1;
select * from tx_refund_bill;
select * from test_new1;
123456789101112131415161718192021 | # 在confluent快速搭建kafka connect - [download confluent](https://www.confluent.io/download/) - quick local start - 创建confluent配置目录 ``` mkdir ~/.confluent ``` - 设置confluent环境 ``` export CONFLUENT_HOME=/home/xingwang/service/confluent-5.4.0 export PATH=$CONFLUENT_HOME/bin:$PATH ``` - 安装debezium - [下载](https://www.confluent.io/hub/debezium/debezium-connector-mysql) - 解压后复制到/home/xingwang/service/confluent-5.4.0/share/java - 安装kafka-connect-jdbc - confluent默认带了kafka-connect-jdbc,只需要额外下载mysql-connector-java-5.1.40.jar放到/home/xingwang/service/confluent-5.4.0/share/java/kafka-connect-jdbc就可以了 - start confluent |
|---|
confluent local start
1234567891011 | - log位置 log在/tmp/下 - confluent 管理页面 [http://172.17.228.163:9021/](http://172.17.228.163:9021/) # 配置connect(配置可以直接在http client中执行(.http)) |
|---|
查看connectors
GET http://172.17.228.163:8083/connectors
delete connnector
curl -XDELETE ‘http://172.17.228.163:8083/connectors/debezium’
创建source debezium connector
curl -H “Content-Type:application/json” -XPUT ‘http://172.17.228.163:8083/connectors/debezium/config’ -d ‘ { “connector.class”: “io.debezium.connector.mysql.MySqlConnector”, “tasks.max”: “1”, “database.hostname”: “localhost”, “database.port”: “3306”, “database.user”: “root”, “database.password”: “[email protected]”, “database.server.id”: “19991”, “database.server.name”: “test_0”, “database.whitelist”: “test”, “include.schema.changes”: “false”, “snapshot.mode”: “schema_only”, “snapshot.locking.mode”: “none”, “database.history.kafka.bootstrap.servers”: “localhost:9092”, “database.history.kafka.topic”: “dbhistory”, “decimal.handling.mode”: “string”, “table.whitelist”: “test.tx_refund_bill”, “database.history.store.only.monitored.tables.ddl”:”true”, “database.history.skip.unparseable.ddl”:”true” }’
查看source debezium connector status
GET http://172.17.228.163:8083/connectors/debezium/status
delete connnector
curl -XDELETE ‘http://172.17.228.163:8083/connectors/jdbc-sink’
创建sink jdbc connector
curl -H “Content-Type:application/json” -XPUT ‘http://172.17.228.163:8083/connectors/jdbc-sink/config’ -d ‘ { “connector.class”: “io.confluent.connect.jdbc.JdbcSinkConnector”, “connection.url”: “jdbc:mysql://localhost:3306/test?nullCatalogMeansCurrent=true”, “connection.user”: “root”, “connection.password”: “[email protected]”, “tasks.max”: “1”, “topics”: “test_0.test.tx_refund_bill”, “table.name.format”: “test_new1”,
1234567 | "insert.mode": "upsert", "pk.fields": "id", "pk.mode": "record_value", "transforms": "ExtractField", "transforms.ExtractField.type": "org.apache.kafka.connect.transforms.ExtractField$Value", "transforms.ExtractField.field": "after" }' |
|---|
查看connectors status
GET http://172.17.228.163:8083/connectors/jdbc-sink/status
```
实验
- 在tx_refund_bill表中insert数据,观察test_new1的变化
- 在tx_refund_bill表中执行update语句,观察test_new1的变化
reference
- confluent doc
- Kafka连接器深度解读之JDBC源连接器
- kafka-jdbc-connector-sink实现kafka中的数据同步到mysql
- Mysql Sink : unknown table X in information_schema Exception
- Kafka Connect JDBC Sink - pk.fields for each topic (table) in one sink configuration
