All Products
Search

This Product

    Realtime Compute for Apache Flink:Version 2021-08-04

    Document Center

    Realtime Compute for Apache Flink:Version 2021-08-04

    Last Updated:Mar 26, 2026

    This page covers what changed in Realtime Compute for Apache Flink in the August 4, 2021 release, including the Ververica Platform (VVP) and Ververica Runtime (VVR) 4.0.8.

    Required actions before upgrading

    Important

    Read this section before you upgrade. These changes require action to avoid job failures or data loss.

    Stopping a job now deletes all its checkpoints. If you need to restore the job's state after stopping, use one of these approaches instead:

    • Suspend and resume — suspending a job preserves state so you can resume from where it stopped.

    • Restore from a savepoint — stop the job, then restore from a previously created savepoint.

    Engine version migration is required for jobs on Flink 1.11 or earlier.

    Starting with VVR 3.0.3, VVP supports running SQL jobs with different engine versions at the same time. The engine version format has also changed to vvr-<VVR version>-flink-<Flink version>.

    Take the following action based on your current engine version:

    Current engine version Action required
    Flink 1.12 Stop and restart the job. VVP automatically updates the engine version to vvr-3.0.3-flink-1.12.
    Flink 1.11 or Flink 1.10 Manually update the engine version to vvr-3.0.3-flink-1.12 or vvr-4.0.8-flink-1.13 before restarting. Restarting without updating causes a timeout error.

    New features

    SQL jobs

    • Fine-grained resource configuration — configure CPU cores and memory size for shared slots in SQL jobs. This lets you right-size resources per job rather than applying a single cluster-wide setting. See Configure a deployment.

    • Version management — publishing a job now generates a development version. Compare versions side by side and roll back to any historical version if a new publish introduces issues. See Manage job versions.

    • Hologres catalog — create a Hologres catalog in the development console to access Hologres tables and read Hologres metadata without leaving Realtime Compute for Apache Flink. See Manage Hologres catalogs.

    • Multi-version SQL jobs — publish SQL jobs using different engine versions in the same workspace. Supported versions: vvr-3.0.3-flink-1.12 and vvr-4.0.8-flink-1.13.

    Connectors

    • MySQL CDC connector — supports full data reading in multiple parallel threads, lock-free algorithms, and resumable upload.

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

    • ApsaraDB RDS for MySQL connector — the Java Database Connectivity (JDBC) driver version can be updated to support more character sets.

    Monitoring and alerting

    • Alert frequency — configure how often alerts fire to reduce noise in high-frequency scenarios.

    • Webhook alerts — send alerts directly to a webhook endpoint for integration with external notification systems.

    Improvements

    • Faster JAR package uploads — the upload method has been optimized, reducing the time required to upload JAR packages.

    • Job names in logs — job names are now included in logs downloaded to Log Service, making it easier to identify which job produced a given log entry.

    Bug fixes

    • Fixed an issue where the currentEmitEventTimeLag metric displayed Log.MAX when some partitions in a Kafka table had no data.