Apache Flink 1.4.1 Released
February 15, 2018 -The Apache Flink community released the first bugfix version of the Apache Flink 1.4 series.
This release includes more than 60 fixes and minor improvements for Flink 1.4.0. The list below includes a detailed list of all fixes.
We highly recommend all users to upgrade to Flink 1.4.1.
Updated Maven dependencies:
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.4.1</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.11</artifactId>
<version>1.4.1</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-clients_2.11</artifactId>
<version>1.4.1</version>
</dependency>
You can find the binaries on the updated Downloads page.
List of resolved issues:
Sub-task
- [FLINK-6321] - RocksDB state backend Checkpointing is not working with KeyedCEP.
- [FLINK-7499] - double buffer release in SpillableSubpartitionView
- [FLINK-7756] - RocksDB state backend Checkpointing (Async and Incremental) is not working with CEP.
- [FLINK-7760] - Restore failing from external checkpointing metadata.
- [FLINK-8323] - Fix Mod scala function bug
Bug
- [FLINK-5506] - Java 8 - CommunityDetection.java:158 - java.lang.NullPointerException
- [FLINK-6951] - Incompatible versions of httpcomponents jars for Flink kinesis connector
- [FLINK-7949] - AsyncWaitOperator is not restarting when queue is full
- [FLINK-8145] - IOManagerAsync not properly shut down in various tests
- [FLINK-8200] - RocksDBAsyncSnapshotTest should use temp fold instead of fold with fixed name
- [FLINK-8226] - Dangling reference generated after NFA clean up timed out SharedBufferEntry
- [FLINK-8230] - NPE in OrcRowInputFormat on nested structs
- [FLINK-8235] - Cannot run spotbugs for single module
- [FLINK-8242] - ClassCastException in OrcTableSource.toOrcPredicate
- [FLINK-8248] - RocksDB state backend Checkpointing is not working with KeyedCEP in 1.4
- [FLINK-8249] - Kinesis Producer didnt configure region
- [FLINK-8261] - Typos in the shading exclusion for jsr305 in the quickstarts
- [FLINK-8263] - Wrong packaging of flink-core in scala quickstarty
- [FLINK-8265] - Missing jackson dependency for flink-mesos
- [FLINK-8270] - TaskManagers do not use correct local path for shipped Keytab files in Yarn deployment modes
- [FLINK-8275] - Flink YARN deployment with Kerberos enabled not working
- [FLINK-8278] - Scala examples in Metric documentation do not compile
- [FLINK-8283] - FlinkKafkaConsumerBase failing on Travis with no output in 10min
- [FLINK-8295] - Netty shading does not work properly
- [FLINK-8306] - FlinkKafkaConsumerBaseTest has invalid mocks on final methods
- [FLINK-8318] - Conflict jackson library with ElasticSearch connector
- [FLINK-8325] - Add COUNT AGG support constant parameter, i.e. COUNT(*), COUNT(1)
- [FLINK-8352] - Flink UI Reports No Error on Job Submission Failures
- [FLINK-8355] - DataSet Should not union a NULL row for AGG without GROUP BY clause.
- [FLINK-8371] - Buffers are not recycled in a non-spilled SpillableSubpartition upon release
- [FLINK-8398] - Stabilize flaky KinesisDataFetcherTests
- [FLINK-8406] - BucketingSink does not detect hadoop file systems
- [FLINK-8409] - Race condition in KafkaConsumerThread leads to potential NPE
- [FLINK-8419] - Kafka consumer's offset metrics are not registered for dynamically discovered partitions
- [FLINK-8421] - HeapInternalTimerService should reconfigure compatible key / namespace serializers on restore
- [FLINK-8433] - Update code example for "Managed Operator State" documentation
- [FLINK-8461] - Wrong logger configurations for shaded Netty
- [FLINK-8466] - ErrorInfo needs to hold Exception as SerializedThrowable
- [FLINK-8484] - Kinesis consumer re-reads closed shards on job restart
- [FLINK-8485] - Running Flink inside Intellij no longer works after upgrading from 1.3.2 to 1.4.0
- [FLINK-8489] - Data is not emitted by second ElasticSearch connector
- [FLINK-8496] - WebUI does not display TM MemorySegment metrics
- [FLINK-8499] - Kryo must not be child-first loaded
- [FLINK-8522] - DefaultOperatorStateBackend writes data in checkpoint that is never read.
- [FLINK-8559] - Exceptions in RocksDBIncrementalSnapshotOperation#takeSnapshot cause job to get stuck
- [FLINK-8561] - SharedBuffer line 573 uses == to compare BufferEntries instead of .equals.
Improvement
- [FLINK-8079] - Skip remaining E2E tests if one failed
- [FLINK-8202] - Update queryable section on configuration page
- [FLINK-8243] - OrcTableSource should recursively read all files in nested directories of the input path.
- [FLINK-8260] - Document API of Kafka 0.11 Producer
- [FLINK-8264] - Add Scala to the parent-first loading patterns
- [FLINK-8271] - upgrade from deprecated classes to AmazonKinesis
- [FLINK-8287] - Flink Kafka Producer docs should clearly state what partitioner is used by default
- [FLINK-8296] - Rework FlinkKafkaConsumerBestTest to not use Java reflection for dependency injection
- [FLINK-8346] - add S3 signature v4 workaround to docs
- [FLINK-8362] - Shade Elasticsearch dependencies away
- [FLINK-8455] - Add Hadoop to the parent-first loading patterns
- [FLINK-8473] - JarListHandler may fail with NPE if directory is deleted
- [FLINK-8571] - Provide an enhanced KeyedStream implementation to use ForwardPartitioner
Test
- [FLINK-8472] - Extend migration tests for Flink 1.4