Apache Flink 1.9.3 Released

April 24, 2020 - Dian Fu (@DianFu11)

The Apache Flink community released the third bugfix version of the Apache Flink 1.9 series.

This release includes 38 fixes and minor improvements for Flink 1.9.2. The list below includes a detailed list of all fixes and improvements.

We highly recommend all users to upgrade to Flink 1.9.3.

Updated Maven dependencies:


You can find the binaries on the updated Downloads page.

List of resolved issues:


  • [FLINK-15143] - Create document for FLIP-49 TM memory model and configuration guide
  • [FLINK-16389] - Bump Kafka 0.10 to


  • [FLINK-11193] - Rocksdb timer service factory configuration option is not settable per job
  • [FLINK-14316] - Stuck in "Job leader ... lost leadership" error
  • [FLINK-14560] - The value of taskmanager.memory.size in flink-conf.yaml is set to zero will cause taskmanager not to work
  • [FLINK-15010] - Temp directories flink-netty-shuffle-* are not cleaned up
  • [FLINK-15085] - HistoryServer dashboard config json out of sync
  • [FLINK-15386] - SingleJobSubmittedJobGraphStore.putJobGraph has a logic error
  • [FLINK-15575] - Azure Filesystem Shades Wrong Package "httpcomponents"
  • [FLINK-15638] - releasing/create_release_branch.sh does not set version in flink-python/pyflink/version.py
  • [FLINK-15812] - HistoryServer archiving is done in Dispatcher main thread
  • [FLINK-15844] - Removal of JobWithJars.buildUserCodeClassLoader method without Configuration breaks backwards compatibility
  • [FLINK-15863] - Fix docs stating that savepoints are relocatable
  • [FLINK-16047] - Blink planner produces wrong aggregate results with state clean up
  • [FLINK-16242] - BinaryGeneric serialization error cause checkpoint failure
  • [FLINK-16308] - SQL connector download links are broken
  • [FLINK-16373] - EmbeddedLeaderService: IllegalStateException: The RPC connection is already closed
  • [FLINK-16573] - Kinesis consumer does not properly shutdown RecordFetcher threads
  • [FLINK-16576] - State inconsistency on restore with memory state backends
  • [FLINK-16696] - Savepoint trigger documentation is insufficient
  • [FLINK-16703] - AkkaRpcActor state machine does not record transition to terminating state.
  • [FLINK-16836] - Losing leadership does not clear rpc connection in JobManagerLeaderListener
  • [FLINK-16860] - Failed to push filter into OrcTableSource when upgrading to 1.9.2
  • [FLINK-16916] - The logic of NullableSerializer#copy is wrong
  • [FLINK-17062] - Fix the conversion from Java row type to Python row type


  • [FLINK-14278] - Pass in ioExecutor into AbstractDispatcherResourceManagerComponentFactory
  • [FLINK-15908] - Add description of support 'pip install' to 1.9.x documents
  • [FLINK-15909] - Add PyPI release process into the subsequent release of 1.9.x
  • [FLINK-15938] - Idle state not cleaned in StreamingJoinOperator and StreamingSemiAntiJoinOperator
  • [FLINK-16018] - Improve error reporting when submitting batch job (instead of AskTimeoutException)
  • [FLINK-16031] - Improve the description in the README file of PyFlink 1.9.x
  • [FLINK-16167] - Update documentation about python shell execution
  • [FLINK-16280] - Fix sample code errors in the documentation about elasticsearch connector
  • [FLINK-16697] - Disable JMX rebinding
  • [FLINK-16862] - Remove example url in quickstarts
  • [FLINK-16942] - ES 5 sink should allow users to select netty transport client


  • [FLINK-11767] - Introduce new TypeSerializerUpgradeTestBase, new PojoSerializerUpgradeTest
  • [FLINK-16454] - Update the copyright year in NOTICE files