这个版本修复了 51 个 bug 和漏洞，并对 Flink 1.14 进行了小的改进。下面是所有 bug 修复和改进的列表(不包括对构建基础结构和构建稳定性的改进)。有关所有更改的完整列表，请参阅: JIRA
我们强烈建议所有用户升级到 Flink 1.14.4。
Release Artifacts Maven Dependencies <dependency> <groupId>org.apache.flink</groupId> <artifactId>flink-java</artifactId> <version>1.14.4</version> </dependency> <dependency> <groupId>org.apache.flink</groupId> <artifactId>flink-streaming-java_2.11</artifactId> <version>1.14.4</version> </dependency> <dependency> <groupId>org.apache.flink</groupId> <artifactId>flink-clients_2.11</artifactId> <version>1.14.4</version> </dependency> Binaries
您可以在更新下载页面上找到二进制文件 Downloads page.
Docker Images
library/flink (official images) apache/flink (ASF repository)
PyPi
apache-flink==1.14.4 Release Notes
Sub-task
[FLINK-21788] - Throw PartitionNotFoundException if the partition file has been lost for blocking shuffle [FLINK-24954] - Reset read buffer request timeout on buffer recycling for sort-shuffle [FLINK-25653] - Move buffer recycle in SortMergeSubpartitionReader out of lock to avoid deadlock [FLINK-25654] - Remove the redundant lock in SortMergeResultPartition [FLINK-25879] - Track used search terms in Matomo [FLINK-25880] - Implement Matomo in Flink documentation
Bug
[FLINK-21752] - NullPointerException on restore in PojoSerializer [FLINK-23946] - Application mode fails fatally when being shut down [FLINK-24334] - Configuration kubernetes.flink.log.dir not working [FLINK-24407] - Pulsar connector chinese document link to Pulsar document location incorrectly. [FLINK-24607] - SourceCoordinator may miss to close SplitEnumerator when failover frequently [FLINK-25171] - When the DDL statement was executed, the column names of the Derived Columns were not validated [FLINK-25199] - StreamEdges are not unique in self-union, which blocks propagation of watermarks [FLINK-25362] - Incorrect dependencies in Table Confluent/Avro docs [FLINK-25407] - Network stack deadlock when cancellation happens during initialisation [FLINK-25466] - TTL configuration could parse in StateTtlConfig#DISABLED [FLINK-25486] - Perjob can not recover from checkpoint when zookeeper leader changes [FLINK-25494] - Duplicate element serializer during DefaultOperatorStateBackendSnapshotStrategy#syncPrepareResources [FLINK-25678] - TaskExecutorStateChangelogStoragesManager.shutdown is not thread-safe [FLINK-25683] - wrong result if table transfrom to DataStream then window process in batch mode [FLINK-25728] - Potential memory leaks in StreamMultipleInputProcessor [FLINK-25732] - Dispatcher#requestMultipleJobDetails returns non-serialiable collection [FLINK-25827] - Potential memory leaks in SourceOperator [FLINK-25856] - Fix use of UserDefinedType in from_elements [FLINK-25883] - The value of DEFAULT_BUNDLE_PROCESSOR_CACHE_SHUTDOWN_THRESHOLD_S is too large [FLINK-25893] - ResourceManagerServiceImpl's lifecycle can lead to exceptions [FLINK-25952] - Savepoint on S3 are not relocatable even if entropy injection is not enabled [FLINK-26039] - Incorrect value getter in map unnest table function [FLINK-26159] - Pulsar Connector: should add description MAX_FETCH_RECORD in doc to explain slow consumption [FLINK-26160] - Pulsar Connector: stopCursor description should be changed. Connector only stop when auto discovery is disabled. [FLINK-26187] - Chinese docs override english aliases [FLINK-26304] - GlobalCommitter can receive failed committables New Feature [FLINK-20188] - Add Documentation for new File Source [FLINK-21407] - Clarify which sources and APIs support which formats Improvement [FLINK-20830] - Add a type of HEADLESS_CLUSTER_IP for rest service type [FLINK-24880] - Error messages "OverflowError: timeout value is too large" shown when executing PyFlink jobs [FLINK-25160] - Make doc clear: tolerable-failed-checkpoints counts consecutive failures [FLINK-25611] - Remove CoordinatorExecutorThreadFactory thread creation guards [FLINK-25650] - Document unaligned checkpoints performance limitations (larger records/flat map/timers/...) [FLINK-25767] - Translation of page 'Working with State' is incomplete [FLINK-25818] - Add explanation how Kafka Source deals with idleness when parallelism is higher then the number of partitions Technical Debt [FLINK-25576] - Update com.h2database:h2 to 2.0.206 [FLINK-25785] - Update com.h2database:h2 to 2.0.210
