Realtime Compute for Apache Flink:August 4, 2021

Major updates in VVP

• After a job is stopped, all checkpoints that are generated for the job are deleted. If you want to restore the status of the job, you can use one of the following methods:

  • Suspend the job and then resume the job.

  • Restore the job from the savepoint after the job is stopped.

• New features

  • Development features

    • Resources can be configured in a fine-grained manner in SQL jobs. This way, the number of CPU cores and the memory size for shared slots can be configured. For more information, see Configure a deployment.

    • When you publish a job, a development version is generated for the job. This way, comparison between versions and rollback to a historical version are supported. For more information, see Manage job versions.

    • A Hologres catalog can be created in the development console of Realtime Compute for Apache Flink. This way, Realtime Compute for Apache Flink can directly access Hologres tables to read the metadata of Hologres by using the Hologres catalog. For more information, see Manage Hologres catalogs.

    • The format of engine versions is changed to vvr-<Ververica Runtime (VVR) version>-flink-<Flink version>.

    • You can publish SQL jobs that use different engine versions. Only the vvr-3.0.3-flink-1.12 and vvr-4.0.8-flink-1.13 versions are supported.

      Important

      In VVR 3.0.3 and later versions, Ververica Platform (VVP) allows you to run SQL jobs that use different engine versions at the same time. The version of the Flink engine that uses VVR 3.0.3 is Flink 1.12. If the engine version of your job is Flink 1.12 or earlier, you can perform the following operations to update the engine version based on the engine version that your job uses:

      • Flink 1.12: Stop and then restart your job. Then, the system automatically updates the engine version of your job to vvr-3.0.3-flink-1.12.

      • Flink 1.11 or Flink 1.10: Manually update the engine version of your job to vvr-3.0.3-flink-1.12 or vvr-4.0.8-flink-1.13, and then restart the job. Otherwise, a timeout error occurs when you start the job.

  • O&M features

    • The alert frequency can be configured.

    • Webhook alerts can be configured.

• Optimized features

  • The method to upload JAR packages is optimized to increase the upload speed.

  • Job names are added to the logs that are downloaded to Log Service. The job names help you identify jobs that have issues.

Major updates in VVR 4.0.8

• Hologres catalogs can be created in the development console of Realtime Compute for Apache Flink.

• Resources can be configured in a fine-grained manner.

• The version of the Java Database Connectivity (JDBC) driver for the ApsaraDB RDS for MySQL connector can be updated to be compatible with more character sets.

• The MySQL Change Data Capture (CDC) connector supports full data reading in multiple parallel threads, lock-free algorithms, and resumable upload.

• The AnalyticDB for MySQL V3.0 connector supports ON DUPLICATE KEY UPDATE semantics.

Major bug fixes in VVR 4.0.8

The issue that Log.MAX is displayed for the currentEmitEventTimeLag metric when some partitions in a Kafka table have no data is fixed.