Apache Flink 1.19.2 Release Announcement

February 12, 2025 - Alexander Fedulov

The Apache Flink Community is pleased to announce the second bug fix release of the Flink 1.19 series.

This release includes 73 bug fixes, vulnerability fixes, and minor improvements for Flink 1.19. Below you will find a list of all bugfixes and improvements (excluding improvements to the build infrastructure and build stability). For a complete list of all changes see: JIRA.

We highly recommend all users upgrade to Flink 1.19.2.

Release Artifacts #

Maven Dependencies #

<dependency>
  <groupId>org.apache.flink</groupId>
  <artifactId>flink-java</artifactId>
  <version>1.19.2</version>
</dependency>
<dependency>
  <groupId>org.apache.flink</groupId>
  <artifactId>flink-streaming-java</artifactId>
  <version>1.19.2</version>
</dependency>
<dependency>
  <groupId>org.apache.flink</groupId>
  <artifactId>flink-clients</artifactId>
  <version>1.19.2</version>
</dependency>

Binaries #

You can find the binaries on the updated Downloads page.

Docker Images #

PyPi #

Release Notes #

    Release Notes - Flink - Version 1.19.2

Bug

  • [FLINK-30899] - FileSystemTableSource with CSV format incorrectly selects fields if filtering for partition
  • [FLINK-33117] - a wrong scala example in udfs page
  • [FLINK-33192] - State memory leak in the Window Operator due to unregistered cleanup timer
  • [FLINK-33571] - Bump json-path from 2.7.0 to 2.9.0
  • [FLINK-33936] - Outputting Identical Results in Mini-Batch Aggregation with Set TTL
  • [FLINK-34252] - WatermarkAssignerOperator should not emit WatermarkStatus.IDLE under continuous data flow
  • [FLINK-35069] - ContinuousProcessingTimeTrigger continuously registers timers in a loop at the end of the window
  • [FLINK-35157] - Sources with watermark alignment get stuck once some subtasks finish
  • [FLINK-35318] - incorrect timezone handling for TIMESTAMP_WITH_LOCAL_TIME_ZONE type during predicate pushdown
  • [FLINK-35334] - Code generation: init method exceeds 64 KB when there is a long array field with Table API
  • [FLINK-35498] - Unexpected argument name conflict error when do extract method params from udf
  • [FLINK-35522] - The source task may get stuck after a failover occurs in batch jobs
  • [FLINK-35531] - Avoid calling hsync in flush method in BaseHadoopFsRecoverableFsDataOutputStream
  • [FLINK-35537] - Error when setting state.backend.rocksdb.compression.per.level
  • [FLINK-35685] - Some metrics in the MetricStore are duplicated when increasing or decreasing task parallelism
  • [FLINK-35699] - Incorrect Jackson shade through fabric8 in flink-kubernetes
  • [FLINK-35721] - I found out that in the Flink SQL documentation it says that Double type cannot be converted to Boolean type, but in reality, it can.
  • [FLINK-35731] - Sink V2 operator is mistakenly assumed always to be parallelism configured
  • [FLINK-35750] - The latency marker metrics aren't updated after failover
  • [FLINK-35764] - TimerGauge is incorrect when update is called during a measurement
  • [FLINK-35786] - NPE in BlobServer / shutdownHook
  • [FLINK-35833] - ArtifactFetchManager always creates artifact dir
  • [FLINK-35885] - proctime aggregate window triggered by watermark
  • [FLINK-35886] - Incorrect watermark idleness timeout accounting when subtask is backpressured/blocked
  • [FLINK-35887] - Null Pointer Exception in TypeExtractor.isRecord when trying to provide type info for interface
  • [FLINK-35935] - CREATE TABLE AS doesn't work with LIMIT
  • [FLINK-35977] - Missing an import in datastream.md
  • [FLINK-36000] - DynamicTableSink#Context's getTargetColumns should not return an array of zero length
  • [FLINK-36116] - Javadocs aren't visible anymore for Flink 1.17 and above
  • [FLINK-36260] - numBytesInLocal and numBuffersInLocal being reported as remote
  • [FLINK-36287] - Sink with topologies should not participate in UC
  • [FLINK-36318] - Fail to restore from 1.18 if LAG function is used
  • [FLINK-36368] - Fix subtask management in CommittableCollector
  • [FLINK-36379] - Improve (Global)Committer with UC disabled
  • [FLINK-36405] - Fix startup issues due to Hive delegation token on kerberos clusters
  • [FLINK-36417] - STATE_TTL doesn't work with WatermarkAssigner
  • [FLINK-36421] - Missing fsync in FsCheckpointStreamFactory
  • [FLINK-36451] - Kubernetes Application JobManager Potential Deadlock and TaskManager Pod Residuals
  • [FLINK-36455] - Sink should commit everything on notifyCheckpointCompleted
  • [FLINK-36511] - FlinkSecurityManager#checkExit StackOverFlow if haltOnSystemExit is enabled
  • [FLINK-36530] - Not able to restore list state from S3
  • [FLINK-36533] - Fix detecting bind failure in case of Netty EPOLL transport
  • [FLINK-36571] - Flink dashboard does not show Busy / Backpressure
  • [FLINK-36788] - Add coverage for GlobalCommitter for SinkV2
  • [FLINK-36941] - Fix Doc for DATE_FORMAT
  • [FLINK-37016] - NPE when ClusterEntrypoing is shut down before initialization
  • [FLINK-37025] - Periodic SQL watermarks can travel back in time
  • [FLINK-37084] - NullPointerException occurs during handling timer of window in PyFlink
  • [FLINK-37168] - TimerRegistrationAction in unregisteredTimers is not cleaned up properly after the timers are registered
  • [FLINK-37183] - Usrlib symlinks are not followed

Improvement

  • [FLINK-33730] - Update the compatibility table to only include last three released versions
  • [FLINK-35353] - Translate "Profiler" page into Chinese
  • [FLINK-35453] - StreamReader Charset fix with UTF8 in core files
  • [FLINK-36593] - Upgrade io.airlift:aircompressor to mitigate CVE
  • [FLINK-36643] - Upgrade aws-java-sdk-core to 1.12.779 to fix vulnerability
  • [FLINK-36689] - [UI] Upgrade runtime web Angular framework and associated deps to latest version
  • [FLINK-36716] - Address vulnerabilities in Flink UI
  • [FLINK-37099] - Document possible Netty4 memory related configurations

Technical Debt