This topic describes the release notes for Realtime Compute for Apache Flink and provides links to relevant references. The release notes provide the major updates and bug fixes in Realtime Compute for Apache Flink in the version that was released on December 3, 2020.
Major updates in VVP
The SQL Preview feature is supported, which allows you to run DML statements in the SQL editor. This feature provides basic debugging capabilities and improves code development efficiency. For more information, see Debug a deployment.
The user-defined connector is supported. If a source or sink is not supported by Realtime Compute for Apache Flink, you can use the user-defined connector to meet your business requirements. For more information, see Manage custom connectors.
The session cluster type is added. You can configure a session cluster to reuse JobManager resources. This reduces resource consumption for jobs that process a small amount of data and accelerates job publishing. This type of clusters is suitable for short-period operations, such as SQL result preview. For more information, see Configure a development and test environment (session cluster).
Major updates in VVR 2.1.2
The time-to-live (TTL) can be configured. You can configure table.exec.state.ttl: 129600000 in the Additional Configuration section of the Advanced tab in the development console of Realtime Compute for Apache Flink. The unit of TTL is milliseconds.
The print connector is supported. You can use this connector to perform online debugging on jobs.
Computed columns can be declared in Message Queue for Apache RocketMQ source tables.
MaxCompute source tables are supported. For more information, see Create an incremental MaxCompute source table.
Major bug fixes in VVR 2.1.2
The issue that a job failover occurs when a field is set to Not Null in the DDL statement of the Hologres connector is fixed.
The issue that data is not delivered if the number of data records that are written to an ApsaraDB RDS for MySQL result table does not exceed the default value of the batchSize parameter is fixed. The default value of this parameter is 5000.
The issue in which the Kafka connector does not call the DeserializationSchema#open() method when a job is restored from a checkpoint or savepoint is fixed. In this issue, the NullPointerException error message appears, and the job fails and cannot be restored.
The issue that an error is returned in a Tablestore result table due to the invalid letter case of a parameter in the WITH clause is fixed.