This topic describes the release notes for fully managed Flink and provides links to the relevant references. The release notes describe the major updates and bug fixes in fully managed Flink in the version that was released on December 12, 2022.
Overview
A new version of fully managed Flink is officially released on December 12, 2022. This version is developed based on Ververica Runtime (VVR) 6.0.3 of Apache Flink 1.15.2. Major defects in the Apache Flink community are fixed in this version. In this version, you can select information in logs to query the related error in the console of fully managed Flink.
If your deployment uses VVR of a version earlier than 6.0.3, we recommend that you upgrade the VVR version to 6.0.3 at the earliest opportunity to ensure stable production.
Features
Errors can be queried based on the selected information in logs. On the Events and Exploration tabs, you can find the log related to an error and select the related information to query the error. This way, you can obtain the cause of and solution to the error.
A non-nullable column of a deployment that uses the CREATE DATABASE AS statement can be created or deleted.
The schema that is inferred by using a Kafka JSON catalog is consistent with the column sequences in a JSON message.
A retry mechanism can be configured for data writing to MaxCompute. This mechanism helps prevent fully managed Flink from failing to perform the COMMIT operation or the createUploadSession operation when service interruption occurs in a Tunnel cluster.
Fully managed Flink is compatible with the invalid syntax for creating a table in the ApsaraDB RDS database. If the syntax is invalid, the Flink Change Data Capture (CDC) task that is running is not affected.
The error message that appears for SQL syntax verification is optimized.
Bug fixes
The following issue is fixed: When the CREATE DATABASE AS statement is executed to synchronize data to Kafka, the setting of the sink.parallelism parameter does not take effect.
The following issue that occurs in VVR 6.0.2 is fixed: The ClassCastException error message appears for a deployment that uses the CREATE DATABASE AS statement.
The following issue that occurs in VVR 6.0.2 is fixed: Curve charts for the latency-related metrics of a MySQL CDC source table are not displayed.
The following issue that occurs in VVR 6.0.2 is fixed: The syntax of SQL hints does not take effect.
The following issue that occurs in VVR 6.0.2 is fixed: A class conflict is reported as a data out-of-range exception during syntax checks.
The following issue that occurs in VVR 6.0.2 is fixed: Server IDs conflict when the CREATE TABLE AS statement is executed in a deployment to write data from one table to multiple sinks.
The following issue is fixed: A JobManager deadlock occurs due to the leader election in a Kubernetes cluster.
The error "Decrease reference to an unknown file" that occurs in the state backend storage GeminiStateBackend is fixed.