- 03 11月, 2017 24 次提交
-
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
This also fixes minor warnings (unchecked casts) and moves the constants into scopes that avoid bridge methods for access outside of the nested classes.
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
The FSDataInputStreamWrapper comes from a time when Flink's FsDataInputStream was not position aware. Not that it is, the FSDataInputStreamWrapper is not required to track its own position, but can simply delegate these calls to the FsDataInputStream. This also adds missing @Override tags.
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
This removes all dependencies on Scala-dependent projects. This commit introduces a hard wired test dependency to 'flink-test-utils_2.11' to avoid introducing a Scala version dependency due to a non-exported test utility.
-
由 Stephan Ewen 提交于
-
由 zentol 提交于
This closes #4931
-
由 Aljoscha Krettek 提交于
Before, we would try and dynamicall load Avro-related classes in several places. Now, we only reflectively instantiate the right AvroUtils and all other operations are methods on this. The default AvroUtils throw exceptions with a helpful message for most operations.
-
由 Aljoscha Krettek 提交于
This also adds a new test that verifies that we correctly register Avro Serializers when they are present and modifies an existing test to verify that we correctly register dummy classes.
-
由 Stephan Ewen 提交于
Excluding Avro in Hadoop will make basic classes like sequence file writers fail to load, even if Avro is not used. Shading Avro is not possible because it is part of the API for users that use Avro types.
-
由 twalthr 提交于
-
由 Stephan Ewen 提交于
-
由 Stephan Ewen 提交于
Moves the SerializationSchema and its related from flink-streaming-java to flink-core. That helps API level projects that depend on those classes to not pull in a dependency on runtime classes, and to not be Scala version dependent.
-
由 Fabian Hueske 提交于
-
由 Fabian Hueske 提交于
[FLINK-7421] [table] Fix serializability of AvroRowSerializationSchema + AvroRowDeserializationSchema.
-
由 Fabian Hueske 提交于
-
由 Stephan Ewen 提交于
These core flink utils are independent of any other runtime classes and are also used both in flink-runtime and in flink-queryable-state (which duplicated the code).
-
由 Stephan Ewen 提交于
The project 'flink-shaded-curator-recipes' was renamed to 'flink-shaded-curator', but some references were not properly updated. This fixes the remaining references.
-
由 Stephan Ewen 提交于
-
由 Till Rohrmann 提交于
-
由 Till Rohrmann 提交于
The LocationPreferenceConstraint defines whether all or any preferred locations have to be taken into consideration when scheduling tasks. Especially for batch jobs where we do lazy scheduling not all input locations might be known for a consumer task. Therefore, we set the location preference constraint to any which means that only those location are taken into consideration which are known at scheduling time. [FLINK-7153] Add test cases Replace AtomicReference with AtomicReferenceUpdater Fix Use static imports Preconditions.checkNotNull Initialize ANY array with number of returned futures Revert formatting changes in Scheduler Set flink-runtime log level in log4j-test.properties to OFF Revert changes to ExecutionVertex#getPreferredLocationsBasedOnInputs Fix failing FailoverRegionTest This closes #4916.
-
由 Till Rohrmann 提交于
-
- 02 11月, 2017 16 次提交
-
-
由 Stephan Ewen 提交于
- Rename the 'flink-shaded-curator-recipes' module to 'flink-shaded-curator', because it actually contains more curator code than just the recipes. - Move the exception handling logic of 'ZooKeeperAccess' directly into the ZooKeeperStateHandleStore
-
由 zentol 提交于
This closes #4927
-
由 Stephan Ewen 提交于
Shading the ZooKeeper dependency makes sure that this specific version of ZooKeeper is used by the Flink runtime module. The ZooKeeper version is sensitive, because we depend on bug fixes in later ZooKeeper versions for Flink's high availability. This prevents situations where for example a set of added dependencies (for example transtive dependencies of Hadoop) cause a different ZooKeeper version to be in the classpath and be loaded. This commit also removes the 'flink-shaded-curator' module, which was originally created to shade guava within curator, but is now obsolete, because newer versions of curator shade guava already.
-
由 Xingcan Cui 提交于
-
由 Till Rohrmann 提交于
Instantiate a HdfsConfiguration as the base configuration when calling HadoopUtils#getHadoopConfiguration() to load the hdfs-site.xml and hdfs-default.xml from the classpath before trying to load an explicitly specified configuration.
-
由 Till Rohrmann 提交于
[FLINK-7960] [tests] Fix race conditions in ExecutionGraphRestartTest#completeCancellingForAllVertices One race condition is between waitUntilJobStatus(eg, JobStatus.FAILING, 1000) and the subsequent completeCancellingForAllVertices where not all execution are in state CANCELLING. The other race condition is between completeCancellingForAllVertices and the fixed delay restart without a delay. The problem is that the 10th task could have failed. In order to restart we would have to complete the cancel for the first 9 tasks. This is enough for the restart strategy to restart the job. If this happens before completeCancellingForAllVertices has also cancelled the execution of the 10th task, it could happen that we cancel a fresh execution. [hotfix] Make WaitForTasks using an AtomicInteger [hotfix] Set optCancelCondition to Optional.empty() in SimpleAckingTaskManagerGateway Add assertion message to ExecutionGraphTestUtils#switchToRunning This closes #4933.
-
由 gyao 提交于
This closes #4914
-
由 gyao 提交于
This closes #4913
-
由 gyao 提交于
This closes #4912
-
由 Tzu-Li (Gordon) Tai 提交于
The affected tests are actually integration tests that spawn a Kafka broker. Renaming it to *ITCase would be more appropriate.
-
由 Piotr Nowojski 提交于
Previously on abort(...) producers were returned to the pool. This was minor bug, probably without any negative side effect, however this patch fixes it and adds additional sanity checks to guard against similar bugs in the future. This closes #4915.
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
initializeState is called before open and since both of those functions relay on chosen semantic, that means checkpointing enable check should happen in initializeState.
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-
由 Piotr Nowojski 提交于
-