| 0700da2 | Tzu-Li (Gordon) Tai | 27 February 2018, 07:03:32 UTC | Commit for release 1.4.2 | 27 February 2018, 07:03:32 UTC |
| b74c705 | Aljoscha Krettek | 22 February 2018, 16:24:33 UTC | [FLINK-8543] Don't call super.close() in AvroKeyValueSinkWriter The call to keyValueWriter.close() in AvroKeyValueSinkWriter.close() will eventually call flush() on the wrapped stream which fails if we close it before(). Now we call flush ourselves before closing the KeyValyeWriter, which internally closes the wrapped stream eventually. | 26 February 2018, 20:12:02 UTC |
| a0193f1 | Matrix42 | 24 February 2018, 13:52:44 UTC | [FLINK-8772] [kafka] Fix missing log parameter This closes #5574. | 26 February 2018, 15:17:04 UTC |
| 0396fc8 | Tzu-Li (Gordon) Tai | 26 February 2018, 11:01:32 UTC | [hotfix] [test] Also trap INT signal in Kafka end-to-end test This allows the test to perform the cleanup procedure (as well as printing any error logs) if an interruption occurred while waiting for the test data to be written to Kafka, therefore increasing visibility of reasons to why the test was stalling. This closes #5568. | 26 February 2018, 15:16:57 UTC |
| 59169d0 | Tzu-Li (Gordon) Tai | 26 February 2018, 08:11:04 UTC | [hotfix] [test] Make test-streaming-kafka010.sh more flexible for local execution | 26 February 2018, 15:16:52 UTC |
| cb6a8a6 | Tzu-Li (Gordon) Tai | 23 February 2018, 11:25:06 UTC | [FLINK-8741] [kafka] Fix incorrect user code classloader in FlinkKafkaConsumer This commit fixes incorrectly using the parent of the user code class loader. Since Kafka 010 / 011 versions directly reuse 09 code, this fix fixes the issue for all versions. This commit also extends the Kafka010Example, so that is uses a custom watermark assigner. This allows our end-to-end tests to have caught this bug. | 26 February 2018, 15:16:45 UTC |
| 0d89a1c | Timo Walther | 23 February 2018, 16:31:44 UTC | [hotfix] Update docs version to Flink 1.4.1 | 23 February 2018, 16:31:44 UTC |
| f06ec38 | Aljoscha Krettek | 21 February 2018, 17:10:55 UTC | [FLINK-8735] Add new StatefulJobSavepointMigrationITCase This new test does not pretend to use legacy state but now instead uses the more modern operator state varieties. | 22 February 2018, 14:38:14 UTC |
| 82e6f8d | Aljoscha Krettek | 21 February 2018, 17:08:13 UTC | [FLINK-8735] Rename StatefulJobSavepointMigrationITCase This is preparation for modifying a new ITCase to use modern state features. | 21 February 2018, 18:14:05 UTC |
| a7df424 | zentol | 07 February 2018, 08:52:23 UTC | [FLINK-8574][travis] Add timestamp to logging messages This closes #5419. | 20 February 2018, 15:48:46 UTC |
| 527faf6 | zentol | 13 February 2018, 09:33:31 UTC | [hotfix][prometheus][tests] Add utility for generating port ranges | 20 February 2018, 15:48:42 UTC |
| d88f43b | zentol | 13 February 2018, 09:33:03 UTC | [hotfix][prometheus] Document internal usage of CollectorRegistry.defaultRegistry It appeared as if the HTTPServer wasn't actually doing anything, but it internally accessed the singleton registry that we also access to register metrics. | 20 February 2018, 15:48:37 UTC |
| 45efe47 | zentol | 13 February 2018, 09:31:31 UTC | [FLINK-8621][prometheus][tests] Remove endpointIsUnavailableAfterReporterIsClosed() The test is inherently unstable as it will always fail if any other server is started on the port between the closing of the reporter and the polling of metrics. This closes #5473. | 20 February 2018, 15:48:25 UTC |
| 528317c | zentol | 18 February 2018, 20:54:56 UTC | [FLINK-8692][docs] Remove extra parenthesis in scala code samples | 18 February 2018, 20:55:59 UTC |
| cc76c32 | Ufuk Celebi | 15 February 2018, 10:58:31 UTC | [FLINK-8303] [docs] Allow to overwrite ruby/gem binary This closes #5395. | 16 February 2018, 08:49:37 UTC |
| d4435e1 | Steven Langbroek | 01 February 2018, 16:52:24 UTC | [FLINK-8303] Add hawkins back to Gemfile | 16 February 2018, 08:49:28 UTC |
| 50b6484 | Steven Langbroek | 31 January 2018, 16:26:28 UTC | [FLINK-8308] Remove explicit yajl-ruby dependency, update Jekyll to 3+ | 16 February 2018, 08:49:14 UTC |
| 054af99 | zentol | 13 February 2018, 11:10:22 UTC | [FLINK-8520][cassandra] Fix race condition This closes #5474. | 16 February 2018, 08:03:35 UTC |
| 1b70f50 | zentol | 07 February 2018, 09:31:41 UTC | [FLINK-8576][QS] Reduce verbosity when classes can't be found This closes #5420. | 16 February 2018, 08:03:35 UTC |
| f2b5635 | Aljoscha Krettek | 15 February 2018, 09:20:58 UTC | [hotfix] Remove more checkState() calls from SharedBuffer serialization | 15 February 2018, 09:20:58 UTC |
| 59f9ded | Fabian Hueske | 14 February 2018, 14:43:16 UTC | [FLINK-8652] [QS] Reduce log level in getKvState to DEBUG. This closes #5489. | 14 February 2018, 19:15:17 UTC |
| a044d9d | Aljoscha Krettek | 14 February 2018, 16:46:51 UTC | [hotfix] Remove costly logging statements from CEP SharedBuffer | 14 February 2018, 16:46:51 UTC |
| bafb91e | zhangminglei | 13 February 2018, 02:33:04 UTC | [FLINK-8423] OperatorChain#pushToOperator catch block may fail with NPE This closes #5447. | 13 February 2018, 08:53:45 UTC |
| beff62d | Stefan Richter | 09 February 2018, 10:30:37 UTC | [FLINK-8571] [DataStream] Introduce utility function that reinterprets a data stream as keyed stream (backport from 1.5 branch) This closes #5439. | 09 February 2018, 16:36:20 UTC |
| 33ebc85 | Tzu-Li (Gordon) Tai | 05 February 2018, 13:16:56 UTC | [FLINK-8362] [elasticsearch] Further improvements for Elasticsearch connector shading - Do not shade Elasticsearch dependencies - Do not shade Flink Elasticseach Connector classes - Also shade log4j-api dependency in Elasticsearch 5 connector. This is required for the log4j-to-slf4j bridge adapter to work properly. - Add NOTICE files for license statements for all ES connectors This closes #5426. This closes #5243. | 08 February 2018, 16:21:23 UTC |
| 0c53e79 | Nico Kruber | 04 January 2018, 20:09:09 UTC | [FLINK-8362][elasticsearch] shade all dependencies | 08 February 2018, 16:21:15 UTC |
| 5f9e367 | Aljoscha Krettek | 07 February 2018, 16:33:03 UTC | [FLINK-7760] Fix deserialization of NFA state in CEP library Before, the condition was being read via in.read() and not in.readFully() | 07 February 2018, 19:02:34 UTC |
| 82f3957 | Tzu-Li (Gordon) Tai | 02 February 2018, 14:53:13 UTC | [FLINK-8275] [security, yarn] Remove test-specific code path in YarnTaskManagerRunner Previously, the YarnTaskManagerRunner contained a code path that exists for the sole purpose of injecting mock runners. Having code paths just to utilize tests in production code is in general a bad idea. This commit fixes this be making YarnTaskManagerRunner a factory-like class, which creates a Runner that contains all the runner’s properties, such as configuration. Unit tests can than test against the contained configuration in the created Runner to validate that everything is configured properly. This closes #5172. | 06 February 2018, 17:30:32 UTC |
| 2014665 | Shuyi Chen | 18 December 2017, 07:50:07 UTC | [FLINK-8275] [security, yarn] Fix keytab local path in YarnTaskManagerRunner | 06 February 2018, 17:30:32 UTC |
| fb1e24e | Tzu-Li (Gordon) Tai | 31 January 2018, 10:13:00 UTC | [FLINK-8421] [test] Increase MAX_MEM_STATE_SIZE and Akka framesize in AbstractEventTimeWindowCheckpointingITCase After adding the TypeSerializerConfigSnapshots of timer serializers to the timers snapshots, the size of the timer snapshots have potentially doubled. This caused the AbstractEventTimeWindowCheckpointingITCase to be failing, because the configured max memory state size and Akka framesize were too small. This commit doubles those sizes. This closes #5362. | 06 February 2018, 17:30:32 UTC |
| 71d682e | Tzu-Li (Gordon) Tai | 24 January 2018, 17:09:44 UTC | [FLINK-8421] [DataStream] Make timer serializers reconfigurable on restore Previously, the key and namespace serializers for the HeapInternalTimerService were not reconfigured on restore to be compatible with previously written serializers. This caused an immediate error to restore savepoints in Flink 1.4.0, since in Flink 1.4.0 we changed the base registrations in the Kryo serializer. That change requires serializer reconfiguration. This commit fixes this by writing also the serializer configuration snapshots of the key and namespace serializer into savepoints, and use them to reconfigure the new serializers on rrestore. This improvement also comes along with making the written data for timer service snapshots versioned. Backwards compatibility with previous non-versioned formats is not broken. | 06 February 2018, 17:30:31 UTC |
| 93d514e | Tzu-Li (Gordon) Tai | 24 January 2018, 17:08:52 UTC | [FLINK-8421] [core] Introduce PostVersionedIOReadableWritable | 06 February 2018, 17:11:26 UTC |
| 319d510 | Tzu-Li (Gordon) Tai | 24 January 2018, 16:15:08 UTC | [FLINK-8421] [core] Allow InstantiationUtil.deserializeObject() to use FailureTolerantObjectInputStream This makes it possible to use the FailureTolerantObjectInputStream in InstantiationUtil.deserializeObject. This behaviour is controlled via a isFailureTolerant flag argument. All previous calls to the method remains to have identical behaviour (the ClassloaderObjectInputStream). | 06 February 2018, 16:31:32 UTC |
| 1f81f6a | Tzu-Li (Gordon) Tai | 24 January 2018, 16:08:13 UTC | [FLINK-8421] [DataStream, tests] Add WindowOperator migration test for Kryo-serialized window keys | 06 February 2018, 16:31:32 UTC |
| e41300d | Tzu-Li (Gordon) Tai | 25 January 2018, 16:45:47 UTC | [hotfix] [test] Remove stale savepoint files no longer used by migration tests This includes: - Removing MigrationVersion.v1_1, since compatilbity for 1.1 is no longer supported (and no migration tests still test that) - Remove all 1.1 test savepoint files (which no migration tests still use) - Remove all 1.2 test savepoint files that are no longer in use (e.g., CEPMigrationTest does not test 1.2 restores anymore) | 06 February 2018, 16:31:32 UTC |
| f68fe3e | Tzu-Li (Gordon) Tai | 01 February 2018, 14:27:46 UTC | [FLINK-8472] [tests] Extend AbstractOperatorRestoreTestBases for Flink 1.4 This closes #5364. | 06 February 2018, 16:31:32 UTC |
| 6f4116f | Tzu-Li (Gordon) Tai | 25 January 2018, 16:32:28 UTC | [FLINK-8472] [fs, test] Extend BucketingSinkMigrationTest for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 8d91096 | Tzu-Li (Gordon) Tai | 25 January 2018, 16:28:17 UTC | [FLINK-8472] [fs, test] Extend ContinuousFileProcessingMigrationTest for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| da47bd2 | Tzu-Li (Gordon) Tai | 25 January 2018, 16:23:22 UTC | [FLINK-8472] [kafka, test] Extend FlinkKafkaConsumerBaseMigrationTest for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 8cef788 | Tzu-Li (Gordon) Tai | 25 January 2018, 15:23:17 UTC | [FLINK-8472] [kinesis, test] Extend FlinkKinesisConsumerMigrationTest for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 83b9727 | Tzu-Li (Gordon) Tai | 25 January 2018, 17:37:05 UTC | [FLINK-8472] [DataStream, test] Extend StatefulJobSavepointMigrationITCase for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 2ada856 | Tzu-Li (Gordon) Tai | 25 January 2018, 17:25:45 UTC | [FLINK-8472] [DataStream, test] Refactor StatefulJobSavepointFrom*MigrationITCase to single ITCase This commit refactors the StatefulJobSavepointFrom12MigrationITCase and StatefulJobSavepointFrom13MigrationITCase to a single class, StatefulJobSavepointMigrationITCase. The new ITCase is parameterized to ensure that all previous versions and state backend variants are covered. | 06 February 2018, 16:31:32 UTC |
| 4fc14cc | Tzu-Li (Gordon) Tai | 25 January 2018, 15:20:31 UTC | [FLINK-8472] [scala, test] Extend StatefulJobSavepointMigrationITCase for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 4af0b59 | Tzu-Li (Gordon) Tai | 25 January 2018, 15:08:53 UTC | [FLINK-8472] [DataStream, test] Extend WindowOperatorMigrationTest for Flink 1.4 | 06 February 2018, 16:31:32 UTC |
| 9276ee5 | Tzu-Li (Gordon) Tai | 01 February 2018, 12:05:56 UTC | [hotfix] [kinesis] Use ArrayList for faster shard state updates Previously, the Kinesis Consumer uses LinkedLists as the underlying implementation for subscribedShardsState. This list is accessed on every record, updating a shard's state using a shard state index (i.e., the order of the shard state in the list). With LinkedLists, this access has linear time complexity, and since this operation happens per record, can have a big performance hit on executions with a huge number of Kinesis shards. This commit changes the list implementation to be an ArrayList for constant-time shard state access. The downside is that when new shards are discovered, there will be a costly array re-size. However, since resharding is not expected to happen so often, this is acceptable. | 06 February 2018, 16:31:32 UTC |
| 7884a4f | Tzu-Li (Gordon) Tai | 01 February 2018, 11:53:42 UTC | [FLINK-8484] [kinesis] Use an equivalence wrapper around restored StreamShardMetadatas to bypass out-of-sync shard infos This closes #5337. | 06 February 2018, 16:31:32 UTC |
| f887575 | Tzu-Li (Gordon) Tai | 01 February 2018, 10:11:48 UTC | [hotfix] [kinesis, test] Use real test classes instead of Mockito in FlinkKinesisConsumerMigrationTest | 06 February 2018, 16:31:32 UTC |
| 426cf46 | Philip Luppens | 23 January 2018, 08:00:23 UTC | [FLINK-8484] [kinesis] Synchronize local shard metainfo's on restore Ensure that a metainfo change in the StreamShardMetadata other than `StreamShardMetadata.shardId` or `StreamShardMetadata.streamName` does not result in the shard not being able to be restored. This handles the corner case where a shard might have been closed (ending sequence number set to not-null) since the last savepoint or checkpoint when a job is restarted from a snapshot state. | 06 February 2018, 16:31:32 UTC |
| ca1e525 | Tzu-Li (Gordon) Tai | 10 January 2018, 05:41:49 UTC | [FLINK-8398] [kinesis, tests] Stabilize flaky KinesisDataFetcherTests Prior to this commit, several unit tests in KinesisDataFetcherTest relied on sleeps to wait until a certain operation happens, in order for the test to pass. This commit removes those sleeps and replaces the test behaviours with OneShotLatches. This closes #5268. | 06 February 2018, 16:31:32 UTC |
| 6922e4c | Tzu-Li (Gordon) Tai | 10 January 2018, 02:11:31 UTC | [FLINK-8398] [kinesis, tests] Cleanup confusing implementations in KinesisDataFetcherTest and related classes The previous implementation of the TestableKinesisDataFetcher was confusing in various ways, causing it hard to be re-used for other tests. This commit contains the following various cleaups: - Remove confusing mocks of source context and checkpoint lock. We now allow users of the TestableKinesisDataFetcher to provide a source context, which should provide the checkpoint lock. - Remove override of emitRecordAndUpdateState(). Strictly speaking, that method should be final. It was previously overriden to allow verifying how many records were output by the fetcher. That verification would be better implemented within a mock source context. - Properly parameterize the output type for the TestableKinesisDataFetcher. - Remove use of PowerMockito in KinesisDataFetcherTest. - Use CheckedThreads to properly capture any exceptions in fetcher / consumer threads in unit tests. - Use assertEquals / assertNull instead of assertTrue where-ever appropriate. | 06 February 2018, 16:31:32 UTC |
| 0ad82b3 | Patrick Lucas | 29 January 2018, 12:58:09 UTC | [hotfix][connectors] Fix log format strings Fixes three occurrences of using '%s' instead of '{}' in logging statements. This closes #5380. | 06 February 2018, 16:31:32 UTC |
| 432273a | Tzu-Li (Gordon) Tai | 22 January 2018, 13:14:44 UTC | [FLINK-8419] [kafka] Register metrics for dynamically discovered Kafka partitions This closes #5335. | 06 February 2018, 16:31:32 UTC |
| e5ac4b0 | Tzu-Li (Gordon) Tai | 22 January 2018, 12:36:20 UTC | [FLINK-8419] [kafka] Move consumer metric group registration to FlinkKafkaConsumerBase This commit is a refactor to move the registration of the consumer metric group (user scope "KafkaConsumer") to FlinkKafkaConsumerBase. Previously, the registration was scattered around in Kafka version-specific subclasses. | 06 February 2018, 16:31:32 UTC |
| a951852 | Tzu-Li (Gordon) Tai | 13 January 2018, 12:35:03 UTC | [hotfix] [kafka] Make AbortedReassignmentException a static class | 06 February 2018, 16:31:32 UTC |
| 0816300 | Tzu-Li (Gordon) Tai | 13 January 2018, 12:13:20 UTC | [FLINK-8409] [kafka] Fix offset committing race condition in KafkaConsumerThread This closes #5329. | 06 February 2018, 16:31:32 UTC |
| c932266 | zentol | 05 February 2018, 12:15:29 UTC | [FLINK-8559][RocksDB] Release resources if snapshot operation fails This closes #5412. | 06 February 2018, 08:45:28 UTC |
| 33efaf7 | kkloudas | 05 February 2018, 13:36:53 UTC | [FLINK-8561] [cep] Fix SharedBuffer.removeEdges to use .equals(). | 05 February 2018, 16:56:53 UTC |
| a35b37a | kkloudas | 25 January 2018, 09:59:30 UTC | [hotfix] [cep] SharedBuffer refactoring. | 05 February 2018, 16:56:37 UTC |
| 624131b | Fabian Hueske | 23 January 2018, 21:10:00 UTC | [FLINK-8242] [table] RexProgramExtractor only translates literals of known types. | 05 February 2018, 13:18:16 UTC |
| 19fcd5e | Fabian Hueske | 12 December 2017, 21:38:51 UTC | [FLINK-8242] [orc] Fix predicate translation if literal is not Serializable. This closes #5345. | 05 February 2018, 13:18:16 UTC |
| e5c1261 | Fabian Hueske | 12 December 2017, 11:18:23 UTC | [FLINK-8243] [orc] OrcTableSource reads input path recursively by default. This closes #5344. | 05 February 2018, 13:18:10 UTC |
| dfa050c | zentol | 29 January 2018, 10:46:08 UTC | [FLINK-8489][ES] Prevent side-effects when modifying user-config | 02 February 2018, 14:59:13 UTC |
| 4219572 | Aljoscha Krettek | 30 January 2018, 16:56:45 UTC | Revert "[FLINK-8411] [State Backends] HeapListState#add(null) will wipe out entire list state" This reverts commit df0526172334ec619ee9a5a70006f4ad2f3e2167. I'm reverting because while this is fixing a bug it is changing behaviour that has been in place for a long while which might break existing user programs. | 31 January 2018, 15:04:52 UTC |
| 110b86d | Fabian Hueske | 12 December 2017, 11:09:14 UTC | [FLINK-8230] [orc] Fix NPEs when reading nested columns. - fixes NPEs for null-valued structs, lists, and maps - fixes NPEs for repeating structs, lists, and maps - adds test for deeply nested data with nulls - adds test for columns with repeating values This closes #5373. | 31 January 2018, 13:46:04 UTC |
| 0cc0572 | zentol | 23 January 2018, 16:32:44 UTC | [FLINK-8496][metrics] Create missing "Network" group This closes #5343. | 31 January 2018, 13:27:57 UTC |
| 14aabf8 | Alejandro Alcalde | 15 January 2018, 18:27:55 UTC | [hotfix][docs] Fix Scala code example in Controlling Latency section This closes #5299. | 29 January 2018, 11:08:43 UTC |
| 09edf6a | Till Rohrmann | 25 January 2018, 10:40:33 UTC | [FLINK-8485] [client] Unblock JobSubmissionClientActor#tryToSubmitJob The JobSubmissionClientActor blocked a ActorSystem's dispatcher thread when requesting the BlobServer port from the cluster. This fails when using the FlinkMiniCluster on a single core machine because we set the number of threads to 1. This commit unblocks the JobSubmissionClientActor#tryToSubmitJob method and sets the lower limit of dispatcher threads to 2 when using the FlinkMiniCluster. This closes #5360. | 25 January 2018, 14:26:10 UTC |
| fb088bc | Till Rohrmann | 10 January 2018, 17:53:38 UTC | [FLINK-7949] Add unit test for AsyncWaitOperator recovery with full queue | 25 January 2018, 14:26:09 UTC |
| a2198b0 | Bartłomiej Tartanus | 30 October 2017, 14:39:43 UTC | [FLINK-7949] Make AsyncWaitOperator recoverable also when queue is full Start emitter thread BEFORE filling up the queue of recovered elements. This guarantees that we won't deadlock inserting the recovered elements, because the emitter can already start processing elements. This closes #4924. | 25 January 2018, 14:26:09 UTC |
| 38278eb | Steven Langbroek | 08 January 2018, 16:27:15 UTC | [FLINK-8352] [web-dashboard] Report error on jar submission failure This closes #5264. | 25 January 2018, 11:08:19 UTC |
| eb63809 | Stephan Ewen | 22 January 2018, 12:32:09 UTC | [FLINK-8406] [bucketing sink] Fix proper access of Hadoop File Systems | 24 January 2018, 17:57:02 UTC |
| 4d1ba45 | Stephan Ewen | 23 January 2018, 21:00:06 UTC | [FLINK-8466] [runtime] Make sure ErrorInfo references no user-defined classes. That way, holding on to the ErrorInfo does not prevent class unloading. However, this implies that the ErrorInfo must not hold strong references to any Exception classes. For that reason, the commit pull the "ground truth" exception into a separate fields, so that the ExecutionGraph logic itself can always assume to have the proper ground-truth exception. | 24 January 2018, 17:57:02 UTC |
| a8ea169 | Stephan Ewen | 24 January 2018, 09:02:00 UTC | [hotfix] [build] Converge Kryo dependency Previously, the Kryo dependency was diverging between the flink-core dependency and the chill dependency. [INFO] +- org.apache.flink:flink-java:jar:1.4.0:compile [INFO] | +- org.apache.flink:flink-core:jar:1.4.0:compile [INFO] | | +- com.esotericsoftware.kryo:kryo:jar:2.24.0:compile .... [INFO] +- org.apache.flink:flink-streaming-java_2.11:jar:1.4.0:compile [INFO] | +- (org.apache.flink:flink-core:jar:1.4.0:compile - omitted for duplicate) [INFO] | +- org.apache.flink:flink-runtime_2.11:jar:1.4.0:compile [INFO] | | +- com.twitter:chill_2.11:jar:0.7.4:compile [INFO] | | | +- com.twitter:chill-java:jar:0.7.4:compile [INFO] | | | | \- (com.esotericsoftware.kryo:kryo:jar:2.21:compile - omitted for conflict with 2.24.0) [INFO] | | | \- (com.esotericsoftware.kryo:kryo:jar:2.21:compile - omitted for conflict with 2.24.0) | 24 January 2018, 17:06:14 UTC |
| 15cb057 | Stephan Ewen | 23 January 2018, 18:58:10 UTC | [FLINK-8499] [core] Force Kryo to be parent-first loaded. | 24 January 2018, 17:06:14 UTC |
| da8446e | Stephan Ewen | 23 January 2018, 20:01:36 UTC | [hotfix] [build] Fix diverging snappy versions. This removes the snappy dependency from flink-core, which is no longer needed since we do not have an Avro dependency in flink-core any more. | 24 January 2018, 17:06:13 UTC |
| c701a33 | Greg Hogan | 10 January 2018, 19:07:57 UTC | [hotfix] [build] Print cache info Print the size of the Maven cache copied for each TravisCI job. | 24 January 2018, 09:52:18 UTC |
| 77df20b | zhangminglei | 15 January 2018, 09:37:41 UTC | [FLINK-8433] [doc] Remove ununsed CheckpointedRestoring interface | 24 January 2018, 09:50:29 UTC |
| e4d7ea2 | zentol | 12 December 2017, 13:04:34 UTC | [FLINK-8079][tests] Stop end-to-end test execution after first failure This closes #5156. | 24 January 2018, 09:42:22 UTC |
| 20be204 | zentol | 22 January 2018, 12:29:34 UTC | [FLINK-8473][webUI] Improve error behavior of JarListHandler This closes #5331. | 24 January 2018, 09:42:16 UTC |
| df05261 | Bowen Li | 12 January 2018, 18:56:42 UTC | [FLINK-8411] [State Backends] HeapListState#add(null) will wipe out entire list state | 23 January 2018, 13:42:16 UTC |
| aa5e491 | 金竹 | 03 January 2018, 15:13:49 UTC | [FLINK-8325] [table] Fix COUNT(*) and COUNT(1). This closes #5241. | 21 January 2018, 09:05:44 UTC |
| bfe3ff0 | 金竹 | 03 January 2018, 15:13:49 UTC | [FLINK-8355] [table] Remove DataSetAggregateWithNullValuesRule. This closes #5320. | 19 January 2018, 21:41:08 UTC |
| 5375522 | Stephan Ewen | 19 January 2018, 12:53:08 UTC | [FLINK-8461] [build] Adjust logger configurations for shaded Netty classnames | 19 January 2018, 12:54:23 UTC |
| 0677f08 | Stephan Ewen | 18 January 2018, 16:57:10 UTC | [FLINK-8455] [core] Make 'org.apache.hadoop.' a 'parent-first' classloading pattern. This change avoid duplication of Hadoop classes between the Flink runtime and the user code. Hadoop (and transitively its dependencies) should be part of the application class loader. The user code classloader is allowed to duplicate transitive dependencies, but not Hadoop's classes directly. This also adds tests to validate parent-first classloading patterns. | 19 January 2018, 08:32:08 UTC |
| 0deaee8 | Greg Hogan | 12 January 2018, 14:29:28 UTC | [hotfix] [docs] Fix typos This closes #5289. | 17 January 2018, 13:44:49 UTC |
| c91544d | Tzu-Li (Gordon) Tai | 20 December 2017, 19:54:40 UTC | [hotfix] [kafka] Remove stale comment on publishing procedures of AbstractFetcher The previous comment mentioned "only now will the fetcher return at least the restored offsets when calling snapshotCurrentState()". This is a remnant of the previous fetcher initialization behaviour, where in the past the fetcher wasn't directly seeded with restored offsets on instantiation. Since this is no longer true, this commit fixes the stale comment to avoid confusion. | 13 January 2018, 03:34:28 UTC |
| c4bfc7d | Tzu-Li (Gordon) Tai | 12 January 2018, 00:45:32 UTC | [FLINK-8306] [kafka, tests] Fix mock verifications on final method Previously, offset commit behavioural tests relied on verifying on AbstractFetcher::commitInternalOffsetsToKafka(). That method is actually final, and could not be mocked. This commit fixes that by implementing a proper mock AbstractFetcher, which keeps track of the offset commits that go through. This closes #5284. | 13 January 2018, 03:34:20 UTC |
| 968683f | Tzu-Li (Gordon) Tai | 20 December 2017, 00:10:44 UTC | [FLINK-8296] [kafka] Rework FlinkKafkaConsumerBaseTest to not rely on Java reflection Reflection was mainly used to inject mocks into private fields of the FlinkKafkaConsumerBase, without the need to fully execute all operator life cycle methods. This, however, caused the unit tests to be too implementation-specific. This commit reworks the FlinkKafkaConsumerBaseTest to remove test consumer instantiation methods that rely on reflection for dependency injection. All tests now instantiate dummy test consumers normally, and let all tests properly execute all operator life cycle methods regardless of the tested logic. This closes #5188. | 13 January 2018, 03:34:10 UTC |
| b7f8a5a | maqingxiang-it | 11 January 2018, 05:44:04 UTC | [hotfix] Fix typo in AbstractMetricGroup.java This closes #5280. | 13 January 2018, 03:33:50 UTC |
| ae00cad | Alejandro Alcalde | 11 January 2018, 12:49:29 UTC | [hotfix] [doc] Fixed doc typo in DataStream API This closes #5283. This closes #5191. | 13 January 2018, 03:33:42 UTC |
| b607a8b | Bowen Li | 21 June 2017, 22:55:19 UTC | [FLINK-6951] [kinesis] Shade httpcomponents dependency for Kinesis connector This closes #4150. | 13 January 2018, 03:33:34 UTC |
| b32ac3a | Dian Fu | 09 December 2017, 02:55:14 UTC | [FLINK-8226] [cep] Dangling reference generated after NFA clean up timed out SharedBufferEntry This closes #5141 | 12 January 2018, 08:06:01 UTC |
| 71ede39 | Nico Kruber | 05 January 2018, 17:18:35 UTC | [FLINK-8371][network] always recycle Buffers when releasing SpillableSubpartition There were places where Buffer instances were not released upon SpillableSubpartition#release() with a view attached to a non-spilled subpartition: 1) SpillableSubpartition#buffer: SpillableSubpartition#release() delegates the recycling to the view, but SpillableSubpartitionView does not clean up the 'buffers' queue (the recycling was only done by the subpartition if there was no view). 2) SpillableSubpartitionView#nextBuffer: If this field is populated when the subpartition is released, it will neither be given out in subsequent SpillableSubpartitionView#getNextBuffer() calls (there was a short path returning 'null' here), nor was it recycled -> similarly to the PipelinesSubpartition implementation, make SpillableSubpartition#release() always clean up and recycle the buffers -> recycle SpillableSubpartitionView#nextBuffer in SpillableSubpartitionView#releaseAllResources() This closes #5276. | 11 January 2018, 10:27:53 UTC |
| a316989 | Nico Kruber | 23 November 2017, 13:58:21 UTC | [FLINK-7499][io] also let AsynchronousBufferFileWriter#writeBlock() recycle the buffer in case of failures This fixes a double-recycle in SpillableSubpartitionView and also makes sure that even if adding the (asynchronous) write operation fails, the buffer is properly freed in code that did not perform this cleanup. It avoids code duplication of this cleanup and it is also more consistent to take over responsibility of the given buffer even if an exception is thrown. [FLINK-7499][io] complete the idiom of ResultSubpartition#add() taking over ownership of the buffer The buffer will now always be released once and at the right time and the caller must not worry about the buffer release if a called function threw an exception. This closes #5275. | 11 January 2018, 10:27:23 UTC |
| 2c667de | Nico Kruber | 23 November 2017, 12:09:37 UTC | [hotfix] only update buffer statistics in SpillableSubpartition#add() if successful | 10 January 2018, 09:42:37 UTC |
| 9398daf | Nico Kruber | 23 November 2017, 13:59:18 UTC | [hotfix] add some more buffer recycling checks in SpillableSubpartitionTest | 10 January 2018, 09:42:37 UTC |
| 6f50ed2 | zentol | 08 January 2018, 16:25:43 UTC | [FLINK-8320][docs] Clarify that only Java 8 is supported | 08 January 2018, 19:26:17 UTC |
| 951d1b9 | yew1eb | 28 December 2017, 17:15:40 UTC | [hotfix][docs][metrics] Fix Threads.Count metric reference This closes #5213. | 08 January 2018, 19:26:12 UTC |
| 2f6e001 | okumin | 06 January 2018, 13:48:02 UTC | [hotfix][docs] Fix DataStream iterations documentation * Fix a scala example which is using a wrong variable * Remove partitioning descriptions * partitioning parameters are already removed from IterativeStream#closeWith/DateStream#iterate * https://github.com/apache/flink/pull/988 * https://github.com/apache/flink/pull/4655 This closes #5249. | 08 January 2018, 19:26:08 UTC |
| 78e3405 | wenlong.lwl | 05 December 2017, 07:16:32 UTC | [FLINK-8200][tests] Use TemporaryFolder in RocksDBAsyncSnapshotTest This closes #5122. | 08 January 2018, 19:26:02 UTC |
