- 29 1月, 2019 8 次提交
-
-
由 zentol 提交于
-
由 Dawid Wysakowicz 提交于
This closes #7436
-
由 eaglewatcherwb 提交于
This closes #7436
-
由 Andrey Zagrebin 提交于
This closes #7188.
-
由 Dawid Wysakowicz 提交于
Logs from failures show that even though the request never finishes, kafka topics are actually deleted. Therefore this PR adds timeout on delete request. If the request times out we check if the topic actually exists. If not then we continue as if the delete request succeeded.
-
由 Igal Shilman 提交于
During the release of Flink 1.7, the value of serialVersionUID was uptick to 2L (was 1L before) And although the AvroSerializer (along with it's snapshot class) were migrated to the new serialization abstraction (hence free from Java serialization), there were composite serializers that were not migrated and were serialized with Java serialization. This commit manually Java-Deserializes the AvroSerializer to support backwards compatability. This closes #7580.
-
由 Bo WANG 提交于
-
由 Tony Feng 提交于
-
- 28 1月, 2019 7 次提交
-
-
由 Till Rohrmann 提交于
The lifecycle methods should not have a RPC timeout. Instead the implementations now use the RpcUtils#INF_TIMEOUT as the timeout.
-
由 Till Rohrmann 提交于
For a better separation of concerns in the JobManagerRunner, this commit introduces a JobMasterService which only exposes the JobMaster's lifecycle methods to the JobManagerRunner. This allows for an easier substitution when testing the JobManagerRunner. This closes #7563.
-
由 Benchao Li 提交于
-
由 leesf 提交于
-
由 Yun Tang 提交于
This closes #7544.
-
由 Aljoscha Krettek 提交于
We add this with the hope of finding clues about FLINK-9920.
-
由 Dawid Wysakowicz 提交于
-
- 26 1月, 2019 5 次提交
-
-
由 Gary Yao 提交于
-
由 Gary Yao 提交于
This closes #7486.
-
由 tison 提交于
-
由 Andrey Zagrebin 提交于
Test non concurrent access of StateSnapshotTransformer Refactor out testNonConcurrentSnapshotTransformerAccess to separte StateSnapshotTransformerTest use element serializer from new meta info, duplicate it in rocksdb transformer factory, test concurrent access for element serializer This closes #7320.
-
由 Gary Yao 提交于
Port YARNSessionCapacitySchedulerITCase#testTaskManagerFailure to flip6 codebase: * Remove assertions that rely on log messages only * Move part where TMs are killed to YARNHighAvailabilityITCase * Rename test to a proper name that describes what it does * Add Javadoc explaning what this test does [FLINK-11390][tests] Move comment to right position [FLINK-11390][tests] Reuse YarnClient from super class Move waitUntilCondition to YarnTestBase [FLINK-11390][tests] Extract method parse hostname Extract method getOnlyApplicationReport Extract method submitJob Extract method getNumberOfSlotsPerTaskManager Extract method getFlinkConfigFromRestApi Delete useless comment Rename: runner -> yarnSessionClusterRunner Delete useless sleep & refactor Reorder methods and add static keyword where possible This closes #7546.
-
- 25 1月, 2019 7 次提交
-
-
由 klion26 提交于
Summary: This patch add the ability about uploading state files using multithread when snapshoting. Differential Revision: https://aone.alibaba-inc.com/code/D818724 This closes #7351.
-
由 Till Rohrmann 提交于
- Moved JobManagerHARecoveryTest#testJobRecoveryWhenLosingLeadership to DispatcherHATest#testJobRecoveryWhenChangingLeadership - Moved JobManagerHARecoveryTest#testFailingJobRecovery to DispatcherHATest#testFailingRecoveryIsAFatalError This closes #7539.
-
由 Till Rohrmann 提交于
The builder creates a TestingHighAvailabilityServices with the standalone implementation of the individual services.
-
由 Till Rohrmann 提交于
-
由 Till Rohrmann 提交于
-
由 Till Rohrmann 提交于
-
由 Till Rohrmann 提交于
-
- 24 1月, 2019 8 次提交
-
-
由 Till Rohrmann 提交于
-
由 Till Rohrmann 提交于
This commit slightly refactors and cleans up the handling of over sized messages. This closes #6876.
-
由 yanghua 提交于
-
由 Till Rohrmann 提交于
Removed testClientNonDetachedListeningBehaviour because the test was only relevant for the JobClientActor. Moved testJobPersistencyWhenJobManagerShutdown to DispatcherHATest#testPersistedJobGraphWhenDispatcherIsShutDown. This closes #7526.
-
由 Till Rohrmann 提交于
- Moved JobManagerStartupTest#testStartupWithPortInUse to BootstrapToolsTest#testActorSystemInstantiationFailureWhenPortOccupied - Moved JobManagerStartupTest#testJobManagerStartupFails to BlobServerTest#testFailureIfStorageDirectoryCannotBeCreated This closes #7541.
-
由 Till Rohrmann 提交于
-
由 Thomas Weise 提交于
-
由 zhijiang 提交于
[FLINK-11282][network] Merge StreamRecordWriter into RecordWriter
-
- 23 1月, 2019 5 次提交
-
-
由 libenchao 提交于
This closes #7569
-
由 Gary Yao 提交于
Port YARNSessionCapacitySchedulerITCase#testClientStartup to new flip6 codebase. Rename test to testStartYarnSessionClusterInQaTeamQueue. Remove -n command line parameter & argument.
-
由 Till Rohrmann 提交于
The JobManagerProcessReapingTest is no longer relevant since we don't start a ProcessReaper as it used to be the case in the legacy mode.
-
由 Till Rohrmann 提交于
This commit renames JobManagerCleanupITCase into BlobsCleanupITCase and uses the MiniCluster instead of the TestingCluster to execute the tests. This closes #7524.
-
由 Tzu-Li (Gordon) Tai 提交于
This closes #7553.
-