Uploaded image for project: 'Debezium'
  1. Debezium
  2. DBZ-935

Error while parsing JSON column type for MySQL



    • Bug
    • Resolution: Done
    • Major
    • 0.9.0.Beta1
    • 0.8.1.Final
    • mysql-connector
    • None


      Steps to reproduce :

      1. Start the connector with the config -

        "name": "demo-mysql-connector",
        "config": {
          "connector.class": "io.debezium.connector.mysql.MySqlConnector",
          "database.history.kafka.bootstrap.servers": "localhost:9092",
          "database.history.kafka.topic": "debz.schema-changes.demo",
          "database.hostname": "localhost",
          "database.user": "dbarr",
          "database.password": "xxx",
          "database.port": "3306",
          "database.server.id": "999",
          "database.server.name": "debz-demo",
          "name": "demo-mysql-connector",
          "database.whitelist": "demo",
          "tasks.max": "1"

      2. Create a table with json column NOT NULL field.

      CREATE TABLE `test_table` (
        `id` bigint(20) unsigned NOT NULL AUTO_INCREMENT,
        `name` varchar(127) NOT NULL,
        `json_col` json NOT NULL,
        PRIMARY KEY (`id`)
      ) ENGINE=InnoDB

      3. Disable mysql strict mode if enabled.

      4. Do inserts

      INSERT INTO `demo`.`test_table` (`id`, `name`) VALUES ('1', 'abc', cast('null' as json));

      No Error. All good.

      INSERT INTO `demo`.`test_table` (`id`, `name`) VALUES ('2', 'def');

      Connector fails. Connector status -

        name: "demo-mysql-connector",
        connector: {
          state: "RUNNING",
          worker_id: ""
        tasks: [
            state: "FAILED",
            trace: "org.apache.kafka.connect.errors.ConnectException: Invalid value: null used for required field: "json_col", schema type: STRING at io.debezium.connector.mysql.AbstractReader.wrap(AbstractReader.java:200) at io.debezium.connector.mysql.AbstractReader.failed(AbstractReader.java:178) at io.debezium.connector.mysql.BinlogReader.handleEvent(BinlogReader.java:451) at com.github.shyiko.mysql.binlog.BinaryLogClient.notifyEventListeners(BinaryLogClient.java:1055) at com.github.shyiko.mysql.binlog.BinaryLogClient.listenForEventPackets(BinaryLogClient.java:913) at com.github.shyiko.mysql.binlog.BinaryLogClient.connect(BinaryLogClient.java:559) at com.github.shyiko.mysql.binlog.BinaryLogClient$7.run(BinaryLogClient.java:793) at java.lang.Thread.run(Thread.java:748) Caused by: org.apache.kafka.connect.errors.DataException: Invalid value: null used for required field: "json_col", schema type: STRING at org.apache.kafka.connect.data.ConnectSchema.validateValue(ConnectSchema.java:220) at org.apache.kafka.connect.data.Struct.validate(Struct.java:233) at org.apache.kafka.connect.data.ConnectSchema.validateValue(ConnectSchema.java:252) at org.apache.kafka.connect.data.Struct.put(Struct.java:216) at org.apache.kafka.connect.data.Struct.put(Struct.java:203) at io.debezium.data.Envelope.create(Envelope.java:262) at io.debezium.connector.mysql.RecordMakers$1.insert(RecordMakers.java:219) at io.debezium.connector.mysql.RecordMakers$RecordsForTable.create(RecordMakers.java:419) at io.debezium.connector.mysql.BinlogReader.handleInsert(BinlogReader.java:765) at io.debezium.connector.mysql.BinlogReader.handleEvent(BinlogReader.java:435) ... 5 more ",
            id: 0,
            worker_id: ""
        type: "source"

      Stacktrace attached.

      This is happening because in the second insert, we don't provide any value for json_col, therefore, mysql inserts an implicit null value for the column. However, this implicit null is different from cast('null' as json). That is why the first insert does not throw any error.

      I have seen the code and I think I know whats happening. Its just that cast('null' as json) produces a null which is handled but the implicit null inserted by mysql in the second insert statement is not handled.

      I will be happy to provide a patch fix for this one.




            deepakbarr Deepak Barr (Inactive)
            deepakbarr Deepak Barr (Inactive)
            0 Vote for this issue
            4 Start watching this issue