使用 Debezium 和 Kafka Connect JDBC Sink 连接器同步数据库时如何重命名主键？

Question

我正在尝试使用 Debezium 将上游数据库中的表同步到下游数据库，遵循 Debezium 博客中描述的方法。

在下游表中，我只需要上游表中的某些列。我还希望更改一些列名（包括主键的名称）。如果我不尝试重命名主键，同步将不会出现任何问题。

我在用：

• 两个数据库均为 SQL Server 2019；和
• Debezium 1.3（但也尝试过 Debezium 1.2，结果相同）。

我在下面列出了我的数据库和连接器设置的完整详细信息。

(1)数据库表定义：

上游表的 DDL 为：

CREATE TABLE [kafkatest.service1].dbo.Users (
    Id int IDENTITY(1,1) NOT NULL,
    Name nvarchar COLLATE SQL_Latin1_General_CP1_CI_AS NULL,
    CONSTRAINT PK_Users PRIMARY KEY (Id)
) GO

下游表的DDL为：

CREATE TABLE [kafkatest.service2].dbo.Users (
    LocalId int IDENTITY(1,1) NOT NULL, // added to avoid IDENTITY_INSERT issue with SQL Server
    ExternalId int NOT NULL,
    ExternalName nvarchar COLLATE SQL_Latin1_General_CP1_CI_AS NULL,
    CONSTRAINT PK_Users PRIMARY KEY (LocalId)
) GO

特别要注意的是，上游表中的“Id”列（主键）应映射到下游表中的“ExternalId”列。

(2)Kafka Connect连接器定义：

源连接器：

{
    "name": "users-connector",
  "config": {
    "connector.class": "io.debezium.connector.sqlserver.SqlServerConnector",
    "tasks.max": "1",
    "database.server.name": "sqlserver",
    "database.hostname": "sqlserver",
    "database.port": "1433",
    "database.user": "sa",
    "database.password": "Password!",
    "database.dbname": "kafkatest.service1",
    "database.history.kafka.bootstrap.servers": "kafka:9092",
    "database.history.kafka.topic": "schema-changes.users",
    "table.whitelist": "dbo.Users"
  }
}

水槽连接器：

{
    "name": "jdbc-sink",
    "config": {
        "connector.class": "io.confluent.connect.jdbc.JdbcSinkConnector",
        "tasks.max": "1",
        "topics.regex": "sqlserver\\.dbo\\.(Users)",
        "connection.url": "jdbc:sqlserver://sqlserver:1433;databaseName=kafkatest.service2",
        "connection.user": "sa",
        "connection.password": "Password!",
        "transforms": "unwrap,route,RenameField",
        "transforms.unwrap.type": "io.debezium.transforms.ExtractNewRecordState",
        "transforms.unwrap.drop.tombstones": "false",
        "transforms.route.type": "org.apache.kafka.connect.transforms.RegexRouter",
        "transforms.route.regex": "(?:[^.]+)\\.(?:[^.]+)\\.([^.]+)",
        "transforms.route.replacement": "$1",
        "transforms.RenameField.type": "org.apache.kafka.connect.transforms.ReplaceField$Value",
        "transforms.RenameField.renames": "Id:ExternalId,Name:ExternalName",
        "auto.create": "false",
        "auto.evolve": "false",
        "insert.mode": "upsert",
        "delete.enabled": "true",
        "pk.fields": "Id",
        "pk.mode": "record_key"
    }
}

据我所知，“pk.mode”需要是“record_key”才能启用删除。我尝试将“pk.fields”值设置为“Id”和“ExternalId”，但都不起作用。

(3) 错误信息：

在第一种情况下（即“pk.fields”：“Id”）我收到以下错误：

2020-08-18 10:16:16,951 INFO   ||  Unable to find fields [SinkRecordField{schema=Schema{INT32}, name='Id', isPrimaryKey=true}] among column names [ExternalId, ExternalName, LocalId]   [io.confluent.connect.jdbc.sink.DbStructure]
2020-08-18 10:16:16,952 ERROR  ||  WorkerSinkTask{id=jdbc-sink-0} Task threw an uncaught and unrecoverable exception. Task is being killed and will not recover until manually restarted. Error: Cannot ALTER TABLE "Users" to add missing field SinkRecordField{schema=Schema{INT32}, name='Id', isPrimaryKey=true}, as the field is not optional and does not have a default value   [org.apache.kafka.connect.runtime.WorkerSinkTask]
org.apache.kafka.connect.errors.ConnectException: Cannot ALTER TABLE "Users" to add missing field SinkRecordField{schema=Schema{INT32}, name='Id', isPrimaryKey=true}, as the field is not optional and does not have a default value

在第二种情况下（即“pk.fields”：“ExternalId”），我收到以下错误：

2020-08-18 10:17:50,192 ERROR  ||  WorkerSinkTask{id=jdbc-sink-0} Task threw an uncaught and unrecoverable exception. Task is being killed and will not recover until manually restarted. Error: PK mode for table 'Users' is RECORD_KEY with configured PK fields [ExternalId], but record key schema does not contain field: ExternalId   [org.apache.kafka.connect.runtime.WorkerSinkTask]
org.apache.kafka.connect.errors.ConnectException: PK mode for table 'Users' is RECORD_KEY with configured PK fields [ExternalId], but record key schema does not contain field: ExternalId

使用 Debezium 时是否可以重命名主键？或者我是否始终需要构建数据库表，以便主键名称在上游和下游数据库中都匹配？

Answer 1

尝试重命名关键字段：

"transforms": "unwrap,route,RenameField,RenameKey",
...
"transforms.RenameKey.type": "org.apache.kafka.connect.transforms.ReplaceField$Key",
"transforms.RenameKey.renames": "Id:ExternalId",

使用时，消息键中的"pk.mode": "record_key"主键用于构建 upsert 查询语句。