460b386 | Dawid Wysakowicz | 22 September 2021, 06:39:40 UTC | Commit for release 1.14.0 | 22 September 2021, 06:39:40 UTC |
a595fbb | Fabian Paul | 21 September 2021, 11:38:15 UTC | [FLINK-24347][connectors/kafka] Keep idle source readers if parallelism is higher than partitions in KafkaSource Before this commit the enumerator signalled the leftover source readers without a partition to finish. This caused that checkpointing was not possible anymore because it is only supported if all tasks are running or FLIP-147 is enabled. This closes #17327 | 21 September 2021, 21:51:36 UTC |
40b0a31 | Chesnay Schepler | 21 September 2021, 13:34:48 UTC | [FLINK-24340] Only print exception on job failure/suspension | 21 September 2021, 13:35:40 UTC |
36c1b5b | Stephan Ewen | 20 September 2021, 00:59:02 UTC | [FLINK-24303][coordination] Failure when creating a source enumerator lead to full failover, not JobManager failure. Instead of letting exceptions during the creation of the Source Enumerator bubble up (and utimately fail the JobManager / Scheduler creation), we now catch those exceptions and trigger a full (global) failover for that case. | 21 September 2021, 08:30:14 UTC |
30d7200 | Dawid Wysakowicz | 07 September 2021, 09:58:18 UTC | [release] Create 1.14 release-notes This closes #17182 | 21 September 2021, 07:35:31 UTC |
69b6e60 | Chesnay Schepler | 09 September 2021, 14:54:16 UTC | [FLINK-24197] Guard against CLRF being split across chunks This commit adds a defense mechanism against https://github.com/netty/netty/issues/11668. If the CRLF prefix of a multipart delimiter is split across 2 chunks an exception is thrown because Netty incorrectly treats CR as data. If CR is the last byte of the current chunk, then we pessimistically assume that this case occurred. We exclude the trailing CR from the current chunk, and add it to the front of the next received chunk. | 20 September 2021, 08:38:12 UTC |
6f2fa53 | Chesnay Schepler | 09 September 2021, 14:51:00 UTC | [hotfix][test] Derive expected files from getFilesToUpload This simplifies debugging if the number of files is modified. | 20 September 2021, 08:38:02 UTC |
026b0cb | Nico Kruber | 17 September 2021, 15:30:22 UTC | [hotfix][docs] fix link from upsert-kafka to kafka Table(!) connector | 20 September 2021, 08:33:27 UTC |
cc19997 | Qingsheng Ren | 14 September 2021, 10:17:56 UTC | [FLINK-24277][connector/kafka] Add OffsetsInitializerValidator interface for validating offset initializer in KafkaSourceBuilder | 18 September 2021, 13:44:00 UTC |
984cfe9 | Qingsheng Ren | 14 September 2021, 07:24:10 UTC | [FLINK-24277][connector/kafka] Remove auto-generated group id in Kafka table source | 18 September 2021, 13:44:00 UTC |
677caa8 | Qingsheng Ren | 14 September 2021, 07:22:00 UTC | [FLINK-24277][connector/kafka] Add configuration for committing offset on checkpoint and disable it if group ID is not specified | 18 September 2021, 13:44:00 UTC |
791c1b9 | Dawid Wysakowicz | 16 September 2021, 12:12:04 UTC | [FLINK-24300] SourceOperator#getAvailableFuture reuses future Callers of SourceOperator#getAvailableFuture might call the method multiple times even if the returned future does not complete. Before the commit each we were creating a new combined future from the SourceReader#isAvailable and the forcedStop one. If the underlying SourceReader#isAvailable has not changed this operation is unnecessary. What is even worse each such operation adds another entry onto the source reader's availability future stack which caused performance regression. The commit reuses the combined future if the underlying SourceReader#isAvailable future has not changed. This closes #17303 | 17 September 2021, 19:19:42 UTC |
63fa7d0 | Anton Kalashnikov | 10 September 2021, 10:13:45 UTC | [FLINK-24233][runtime] Ignore message about new buffer size if the reader doesn't ready yet | 17 September 2021, 18:16:54 UTC |
64902ea | David Moravek | 06 September 2021, 09:39:24 UTC | [FLINK-24129][connectors-pulsar] Harden TopicRangeTest.rangeCreationHaveALimitedScope. | 17 September 2021, 17:56:29 UTC |
a2b612a | syhily | 16 September 2021, 17:13:52 UTC | [FLINK-23969][connector/pulsar] Create e2e tests for pulsar connector. (cherry picked from commit b6086203d5fc0a08a330dd0069fbe1359ceac97a) | 17 September 2021, 14:32:29 UTC |
d818ccb | Fabian Paul | 15 September 2021, 14:49:16 UTC | [hotfix][connectors/kafka] Rename EventDeSerializer to EventDeSerializationSchema in examples | 17 September 2021, 13:30:56 UTC |
06d4828 | Fabian Paul | 15 September 2021, 10:46:41 UTC | [FLINK-24292][connectors/kafka] Use KafkaSink in examples instead of FlinkKafkaProducer | 17 September 2021, 13:30:56 UTC |
5dd99ed | Fabian Paul | 14 September 2021, 14:04:25 UTC | [FLINK-24281][connectors/kafka] Migrate all format tests from FlinkKafkaProducer to KafkaSink | 17 September 2021, 13:19:38 UTC |
83fd46f | Fabian Paul | 15 September 2021, 12:56:56 UTC | [FLINK-24281][connectors/kafka] Only allow KafkaSinkBuilder creation with KafkaSink.builder() | 17 September 2021, 13:19:38 UTC |
587e677 | Fabian Paul | 14 September 2021, 12:29:07 UTC | [hotfix][connectors/kafka] Remove unused code from KafkaDynamicSink | 17 September 2021, 13:19:38 UTC |
c05dd39 | huangxingbo | 17 September 2021, 08:12:21 UTC | [FLINK-24317][python][tests] Optimize the implementation of Top2 in test_flat_aggregate This closes #17309. | 17 September 2021, 11:13:32 UTC |
9c4262e | Ingo Bürk | 16 September 2021, 12:00:29 UTC | [hotfix] Update frontend-maven-plugin This backports part of a commit to update the frontend-maven-plugin. Doing so prevents Angular's compatibility compiler output to be logged on ERROR level. | 17 September 2021, 09:58:36 UTC |
1ad53cf | Dawid Wysakowicz | 16 September 2021, 06:58:50 UTC | [hotfix] Fix a typo in 'estimatedTimeToConsume(r)BuffersMs' | 16 September 2021, 12:20:25 UTC |
4cb7624 | Anton Kalashnikov | 14 September 2021, 15:33:46 UTC | [FLINK-24133][core] Network failure test replaced by comment in the code due to high expense of stabilizing such low important test | 16 September 2021, 08:12:02 UTC |
4ebd057 | huangxingbo | 16 September 2021, 07:22:11 UTC | [FLINK-24305][python] Limit the protobuf version<3.18 This closes #17298. | 16 September 2021, 07:40:36 UTC |
8d148a8 | Brian Zhou | 09 September 2021, 03:33:41 UTC | [FLINK-24059][Connectors/Common][test] Allow SourceReaderTestBase.NUM_SPLITS to be overridden (#17064) | 15 September 2021, 12:40:50 UTC |
86f96cd | Timo Walther | 14 September 2021, 09:15:07 UTC | [hotfix][docs] Add glossary entry for 'Table program' | 15 September 2021, 07:49:42 UTC |
02040c9 | Timo Walther | 13 September 2021, 12:07:53 UTC | [FLINK-21589][docs] Document table pipeline upgrades This closes #17260 | 15 September 2021, 07:49:34 UTC |
5f2ec45 | syhily | 14 September 2021, 15:03:40 UTC | [FLINK-24283][connector/pulsar] Use stick key consumer in Key_Shared subscription. This would make sure Pulsar won't treat the flink reader as a shared consumer. This fix https://github.com/apache/pulsar/pull/12035 | 14 September 2021, 19:52:55 UTC |
8c1033d | Fabian Paul | 14 September 2021, 14:26:15 UTC | [FLINK-24282][connectors/kafka] Make topic selector for KafkaSink serializable It is possible to calculate the target topic per record. Therefore users can provide a lambda when constructing the KafkaSink. Before this commit the lambda was not marked as serializable and could not be transferred to the workers. | 14 September 2021, 19:52:22 UTC |
d008496 | Roman Khachatryan | 25 August 2021, 10:16:19 UTC | [FLINK-23607][state/changelog] Document checkpointing.changelog in pom.xml | 14 September 2021, 13:28:31 UTC |
2169b79 | Roman Khachatryan | 25 August 2021, 09:29:05 UTC | [FLINK-23607][state/changelog] Cleanup dependencies of DSTL DFS | 14 September 2021, 13:28:31 UTC |
e7ba160 | Roman Khachatryan | 25 August 2021, 09:13:26 UTC | [FLINK-23607][state/changelog] Make Changelog Backend a transitive dependency ...of flink-test-utils instead of direct dependency of each module. | 14 September 2021, 13:28:31 UTC |
918fde8 | Marios Trivyzas | 13 September 2021, 12:21:38 UTC | [hotfix][docs] Fix output of test job Previously, the output listed here was the one for `head` and not `tail` as it's described in the steps to execute the test. This closes #17261. | 14 September 2021, 07:36:28 UTC |
753947a | Yun Gao | 13 September 2021, 06:39:18 UTC | [FLINK-24266][checkpoint] Log improvement for aborting checkpoint due to tasks are finishing This closes #17250. | 14 September 2021, 07:03:51 UTC |
83ab9eb | Yun Gao | 13 September 2021, 07:26:22 UTC | [hotfix][connector] Use null to replace empty pending transaction in 2pc sink This closes #17251. | 14 September 2021, 07:03:51 UTC |
574e80f | huangxingbo | 14 September 2021, 03:23:01 UTC | [FLINK-24276][python] Avoid confusing output when executing in loopback mode This closes #17274. | 14 September 2021, 06:38:51 UTC |
beb7009 | Dian Fu | 14 September 2021, 03:29:20 UTC | [FLINK-24267][python][docs] Update the example tutorial to use latest API This closes #17273. | 14 September 2021, 06:36:44 UTC |
b26f7e7 | Qingsheng Ren | 23 August 2021, 03:26:44 UTC | [FLINK-23773][connector/kafka] Mark empty splits as finished to cleanup states in SplitFetcher | 14 September 2021, 05:34:48 UTC |
52a26c8 | Qingsheng Ren | 23 August 2021, 04:13:32 UTC | [hotfix][connector/common] Expose splitFinishedHook in SplitFetcher for fine-grained testing | 14 September 2021, 05:34:48 UTC |
da82cb1 | Yuan Mei | 14 September 2021, 03:09:40 UTC | [FLINK-24155][documentation] Sync Chinese version of documentation to configure CheckpointFailureManager (#17173) (#17252) | 14 September 2021, 03:09:40 UTC |
742ba0a | sammieliu | 08 September 2021, 12:20:32 UTC | [FLINK-24212][k8s]fix the problem that kerberos krb5.conf file is mounted as empty directory, not the expected file This closes #17198 | 14 September 2021, 01:58:01 UTC |
28677f3 | syhily | 13 September 2021, 08:40:08 UTC | [FLINK-24206][connector/pulsar] Close the pulsar client properly. Pulsar Consumer would be closed in SplitReader if it reaches the end of split. The registered Consumer would also be closed in PulsarClient if we call PulsarClient.close(). This would cause race condition. We have to use PulsarClient.shutdown() which don't close Consumer instead. This closes #17255 | 14 September 2021, 01:40:05 UTC |
1e3a923 | martijnvisser | 13 September 2021, 14:32:42 UTC | [hotfix][docs] Adding new sources on overview pages This closes #17266 | 13 September 2021, 16:02:17 UTC |
dc94e85 | Dawid Wysakowicz | 03 September 2021, 10:05:03 UTC | [FLINK-24098] Document FLIP-147 capabiliites and limitations This closes #17135 | 13 September 2021, 13:00:40 UTC |
b7e4550 | Dian Fu | 10 September 2021, 09:59:12 UTC | [FLINK-24243][python] Cleanup code to use latest API to avoid warnings This closes #17237. | 13 September 2021, 12:22:23 UTC |
daad7e2 | Anton Kalashnikov | 25 August 2021, 16:38:49 UTC | [FLINK-23458][docs] Added the network buffer documentation along with the buffer debloat doc | 13 September 2021, 10:16:29 UTC |
23d63ed | huangxingbo | 10 September 2021, 11:08:31 UTC | [FLINK-24245][python] Fix the problem caused by multiple jobs sharing the loopback mode address stored in the environment variable in PyFlink This closes #17239. | 13 September 2021, 07:28:54 UTC |
79a801a | Piotr Nowojski | 09 September 2021, 12:22:33 UTC | [FLINK-24184][task] Introduce lock to guard against race conditions around shouldInterruptOnCancel Previously there was a potential race condition in disabling interrupts while closing resources. It used be guarded by a volatile variable, but there might have been a race condition when: 1. interrupter thread first checked the shouldInterruptOnCancel flag 2. shouldInterruptOnCancel flag switched to false as Task/StreamTask entered cleaning up phase 3. interrupter issued an interrupt while Task/StreamTask are closing/releasing resources, potentially causing a memory leak | 13 September 2021, 07:24:41 UTC |
b3a97b1 | Dian Fu | 10 September 2021, 08:56:10 UTC | [FLINK-24244][python] Logging whether it's executed in loopback mode This closes #17234. | 13 September 2021, 03:21:32 UTC |
ed022d2 | Dian Fu | 09 September 2021, 02:53:42 UTC | [FLINK-18880][python] Respect configurations defined in flink-conf.yaml and environment variables when executing in local mode This closes #17216. | 13 September 2021, 03:17:51 UTC |
22e7617 | Yun Gao | 02 September 2021, 11:41:40 UTC | [FLINK-24065][connector] Upgrade the state of TwoPhaseCommitSink to support empty transaction after finished | 13 September 2021, 02:38:55 UTC |
7b213e4 | Stephan Ewen | 27 August 2021, 00:26:27 UTC | [hotfix] Improve language for error message when checkpoint is declined due to some tasks being finished already. | 10 September 2021, 14:08:18 UTC |
b32e6a6 | Yufan Sheng | 10 September 2021, 06:01:34 UTC | [FLINK-23864][docs] Add pulsar connector document (Chinese & English). | 10 September 2021, 12:53:14 UTC |
982db57 | Yufan Sheng | 10 September 2021, 04:10:48 UTC | [FLINK-23864][connector/pulsar] Remove PULSAR_AUTO_UPDATE_PARTITIONS option. | 10 September 2021, 12:53:14 UTC |
d40cfc6 | Yufan Sheng | 10 September 2021, 04:10:12 UTC | [FLINK-23864][connector/pulsar] Release Pulsar Message if user enable poolMessage option. | 10 September 2021, 12:53:14 UTC |
77012db | Yufan Sheng | 10 September 2021, 04:09:54 UTC | [FLINK-23864][docs] Add flink-connector-pulsar module to flink-docs, auto generate the config document. | 10 September 2021, 12:53:14 UTC |
df3b723 | Dawid Wysakowicz | 08 September 2021, 14:41:16 UTC | [FLINK-24161] Fix interplay of stop-with-savepoint w/o drain with final checkpoints This closes #17200 | 10 September 2021, 09:18:24 UTC |
d9e6abf | huangxingbo | 07 September 2021, 09:24:15 UTC | [FLINK-24123][python] Optimize the python operator instances of the same job to share one python environment resources in the same jvm This closes #17180. | 10 September 2021, 06:05:01 UTC |
4e68c45 | Yufan Sheng | 08 September 2021, 15:13:06 UTC | [FLINK-23944][connector/pulsar] Enable PulsarSourceITCase.testTaskManagerFailure after test framework was fixed. This closes #17201 | 10 September 2021, 03:00:18 UTC |
4e732ad | Yufan Sheng | 07 September 2021, 06:51:19 UTC | [FLINK-23848][connector/pulsar] Make sure the topic would only be sought once. This closes #17199 | 10 September 2021, 02:58:42 UTC |
1a7bc49 | Yufan Sheng | 03 September 2021, 07:45:04 UTC | [FLINK-23848][connector/pulsar] Fix the consumer not found. | 10 September 2021, 02:58:41 UTC |
3621b00 | Dian Fu | 09 September 2021, 13:00:56 UTC | [FLINK-23345][python] Limits the version requests to 2.26.0 or above This closes #17219. | 10 September 2021, 02:01:34 UTC |
c56ecf3 | Stephan Ewen | 24 August 2021, 14:06:04 UTC | [FLINK-23527][core] Clarify semantics of SourceFunction.cancel() with respect to thread interruptions. | 09 September 2021, 15:18:44 UTC |
9a26fe9 | Stephan Ewen | 24 August 2021, 13:37:09 UTC | [hotfix][core] Make the example for checkpointed SourceFunction a bit less strange. | 09 September 2021, 15:18:37 UTC |
18e94e4 | Stephan Ewen | 24 August 2021, 13:36:12 UTC | [hotfix] Deduplicate JavaDocs in SourceFunction | 09 September 2021, 15:18:25 UTC |
814f278 | Stephan Ewen | 24 August 2021, 13:22:56 UTC | [hotfix] Remove outdated JavaDocs in SourceFunction about deprecated TimeCharacteristic | 09 September 2021, 15:18:13 UTC |
2af65df | Stephan Ewen | 09 September 2021, 14:14:54 UTC | [hotfix][connectors] Deduplicate config default in FutureCompletingBlockingQueue. | 09 September 2021, 15:17:17 UTC |
a76854f | Stephan Ewen | 19 April 2021, 16:22:07 UTC | [FLINK-22358][connector base] Add stability annotations to connector base and iterator sources. | 09 September 2021, 15:17:04 UTC |
2a13b24 | Stephan Ewen | 09 September 2021, 13:48:19 UTC | [hotfix][core] Improve/fix JavaDocs in SplitEnumeratorContext class. | 09 September 2021, 15:12:27 UTC |
4aa457d | Stephan Ewen | 19 April 2021, 16:05:01 UTC | [FLINK-22357][core] Upgrades API stability of unified source API to @Public (= stable). | 09 September 2021, 15:12:23 UTC |
2459a3c | Chesnay Schepler | 09 September 2021, 07:40:20 UTC | [FLINK-24213][qs] Use single lock in ServerConnection | 09 September 2021, 12:37:13 UTC |
a2185c0 | Chesnay Schepler | 09 September 2021, 07:34:26 UTC | [FLINK-24213][qs] Introduce factory for established connection | 09 September 2021, 12:37:06 UTC |
2db50d6 | Dian Fu | 09 September 2021, 09:27:07 UTC | [FLINK-24137][python] Fix the issue that tests planned to run in process mode was actually executed in loopback mode This closes #17214. | 09 September 2021, 12:25:11 UTC |
4d131d5 | Dian Fu | 09 September 2021, 07:34:17 UTC | [FLINK-24137][python] Fix the cases that Python boot process exits before Python SDK harness connects to the Java Operator in lookback mode This closes #17214. | 09 September 2021, 12:25:07 UTC |
e859af6 | Roman Khachatryan | 07 September 2021, 00:36:45 UTC | [FLINK-22889][tests] Increase timeouts in JdbcExactlyOnceSinkE2eTest ...to tolerate occasional long XA PREPARE executions in MySQL. | 09 September 2021, 08:47:59 UTC |
8e982b7 | LB-Yu | 15 August 2021, 10:45:47 UTC | [FLINK-23462] [docs-zh] Translate the abfs documentation to chinese (cherry picked from commit 19804811351c5a66503bf2c6d3b626d1fd1f8c65) | 09 September 2021, 08:08:27 UTC |
adc9d43 | Chesnay Schepler | 09 September 2021, 07:48:26 UTC | [hotfix] Add .eslintcache to .gitignore The file only exists on master; the entry is just added for convenience. | 09 September 2021, 07:48:28 UTC |
e7766f7 | Chesnay Schepler | 09 September 2021, 07:27:32 UTC | [FLINK-24031] Extend vcs.xml | 09 September 2021, 07:45:37 UTC |
b56041b | Roman Khachatryan | 06 September 2021, 20:13:28 UTC | [FLINK-24160][tests] Use FS checkpoint storage in PartiallyFinishedSourcesITCase This closes #17165. | 08 September 2021, 16:04:17 UTC |
28109d0 | Yun Gao | 03 September 2021, 02:39:59 UTC | [FLINK-24068][checkpoint] Also check for alignment start when received EndOfPartition from pending channels This closes #17131. | 08 September 2021, 15:50:41 UTC |
f907ef2 | martijnvisser | 07 September 2021, 11:55:23 UTC | [FLINK-24034] Upgrade commons-compress to 1.21 | 08 September 2021, 14:24:36 UTC |
21fd895 | Qingsheng Ren | 07 September 2021, 08:16:42 UTC | [FLINK-23807][connector/testing-framework] Use RestClient to detect TaskManager failure in test environments | 08 September 2021, 13:51:55 UTC |
b434f0a | Qingsheng Ren | 02 September 2021, 08:02:00 UTC | [FLINK-24126][connector/kafka] Use increment of bytes consumed/produced for updating numBytesIn/Out in Kafka connector (cherry picked from commit 0dc46ca03933890bc14b9025702b56e32ee8b40f) | 08 September 2021, 13:51:09 UTC |
7818495 | David Moravek | 08 September 2021, 08:41:17 UTC | [FLINK-23961][kubernetes] Hint user about missing 'kubernetes.cluster-id' config option | 08 September 2021, 08:41:55 UTC |
adb9009 | Francesco Guardiani | 08 September 2021, 08:30:30 UTC | [FLINK-24170][build] Use maven-directory-plugin:directory-of goal | 08 September 2021, 08:30:58 UTC |
a4cd43b | Chesnay Schepler | 07 September 2021, 09:56:25 UTC | [FLINK-24183][legal] Add aws-java-sdk-s3 to source NOTICE | 08 September 2021, 08:16:35 UTC |
602a177 | wangfeifan | 06 September 2021, 06:43:52 UTC | [FLINK-23949][checkpoint] Fix first incremental checkpoint after a savepoint degenerate into a full checkpoint this fix #16969. Co-authored-by: wangfeifan <zoltar9264@163.com> Co-authored-by: jinghaihang <jinghaihang_hrbeu@163.com> | 08 September 2021, 07:18:33 UTC |
3a8a6b0 | Yun Gao | 07 September 2021, 05:41:46 UTC | [FLINK-24163][test] Increase the checkpoint timeout for PartiallyFinishedSourcesITCase This closes #17170. | 08 September 2021, 03:12:23 UTC |
c10de14 | Leonard Xu | 08 September 2021, 02:43:22 UTC | [FLINK-24115][docs] Fix outdated SQL Temporal Join example (#17122) | 08 September 2021, 02:43:22 UTC |
6b4abd5 | Dian Fu | 07 September 2021, 12:59:47 UTC | [FLINK-24137][python][tests] Upload logs for PyFlink tests | 08 September 2021, 01:42:37 UTC |
2235dcb | Dian Fu | 07 September 2021, 12:58:39 UTC | [FLINK-24137][python] Add more logs during starting up Python worker | 08 September 2021, 01:42:33 UTC |
6472d5b | Matthias Pohl | 06 September 2021, 13:40:40 UTC | [FLINK-23611][yarn-tests] Disables INFO log messages coming from YARN's ResourceLocalizationService We observed regular INFO log messages being produced by ResourceLocalizationService after the test ran into a timeout: ``` 22:51:31,785 [AsyncDispatcher event handler] INFO org.apache.hadoop.yarn.server.nodemanager.containermanager.localizer.ResourceLocalizationService [] - Cache Size Before Clean: 0, Total Deleted: 0, Public Deleted: 0, Private Deleted: 0 22:51:32,398 [AsyncDispatcher event handler] INFO org.apache.hadoop.yarn.server.nodemanager.containermanager.localizer.ResourceLocalizationService [] - Cache Size Before Clean: 0, Total Deleted: 0, Public Deleted: 0, Private Deleted: 0 ``` These log messages appeared every 10 minutes which prevented the ci/tools/test_controller.sh's watchdog mechanism to kick in. The watchdog mechanism relies on no output being produced for a given amount of time. This way, the test_controller script was unable to archive the YARN Flink log. | 07 September 2021, 14:06:24 UTC |
de07689 | Matthias Pohl | 27 August 2021, 10:00:50 UTC | [hotfix][yarn-tests] Replaces runtime by timestamp The intend is to improve local debugging matching log events in different log files (YARN, Flink) via the timestamp. | 07 September 2021, 14:06:24 UTC |
27843b3 | Matthias Pohl | 25 August 2021, 13:49:57 UTC | [hotfix][yarn-tests] Fixes path definition for local YARN-test working directory artifact collection | 07 September 2021, 14:06:24 UTC |
3c20439 | Roman Khachatryan | 06 September 2021, 17:24:08 UTC | [FLINK-24162][tests] Check MAX_WATERMARK per attempt in FLIP-147 tests Adjust DrainingValidator to validate subtask events if: - finished normally, last attempt - finished normally, recovered, finished - both attempts must emit MAX_WATERMARK and end input Skip if task wasn't finished and the attempt failed (by any subtask). | 07 September 2021, 13:46:40 UTC |
ba749b7 | Roman Khachatryan | 06 September 2021, 17:59:09 UTC | [hotfix][tests] Parameterize PartiallyFinishedSourcesITCase with failover strategy | 07 September 2021, 13:46:40 UTC |
510c2fc | Arvid Heise | 05 September 2021, 21:39:32 UTC | [FLINK-24151][connectors/kafka] Add concurrent checkpoint test to KafkaSinkITCase. | 07 September 2021, 12:07:24 UTC |
869befd | Arvid Heise | 05 September 2021, 20:08:07 UTC | [FLINK-24131][connectors/kafka] Fix KafkaWriter currentSendTime metric. The metric can only be registered once and should simply take the currentProducer to calculate. | 07 September 2021, 12:07:24 UTC |
40da6fc | Arvid Heise | 04 September 2021, 20:56:24 UTC | [FLINK-24131][connectors/kafka] Improve threading model of KafkaWriter. Removed pending records as it doesn't add anything to KafkaWriter#flush (same post-condition as per JavaDoc) but introduces instabilities because of concurrency. | 07 September 2021, 12:07:24 UTC |