- 08 11月, 2018 7 次提交
-
-
由 Stefan Richter 提交于
This closes #7032.
-
由 jrthe42 提交于
This closes #7014.
-
由 alex04.wang 提交于
-
由 Stefan Richter 提交于
[FLINK-10816][cep] Fix LockableTypeSerializer.duplicate() to consider deep duplication of element serializer This closes #7049.
-
由 zentol 提交于
-
由 hequn8128 提交于
This closes #6949
-
由 Till Rohrmann 提交于
-
- 07 11月, 2018 17 次提交
-
-
由 zentol 提交于
-
由 zentol 提交于
The tests attempts to re-bind a server to a specific port. If the port is taken the test will fail. The test is now skipped instead if this happens. This closes #7040.
-
由 Till Rohrmann 提交于
Fix test_confluent_schema_registry.sh and test_sql_client.sh end-to-end tests which failed because of missing arguments for kafka-common.sh and the newly introduce set -e flag. This closes #7034.
-
由 Tzu-Li Chen 提交于
-
由 Timo Walther 提交于
This commit modifies the test infrastructure to allow bash's basic error handling mechanism with set -e. Many tests are not ready for a globally defined strict error handling. For now, at least newly developed tests should consider this flag. If a test causes an error, a test fails with "[FAIL] Test script contains errors". See the end-to-end test README for more information about how to develop tests in the future. This closes #7023.
-
由 zentol 提交于
-
由 zentol 提交于
-
由 yanghua 提交于
-
由 Stefan Richter 提交于
-
由 Stefan Richter 提交于
[FLINK-10353][kafka] Support change of transactional semantics in Kafka Producer with existing state This closes #7010.
-
由 Zhijiang 提交于
-
由 Timo Walther 提交于
-
由 Timo Walther 提交于
-
由 zentol 提交于
-
由 Timo Walther 提交于
-
由 Rong Rong 提交于
This closes #5863.
-
由 Stefan Richter 提交于
[FLINK-10720][tests] Add deployment end-to-end stress test with many inflated task deployment desciptors This closes #6994.
-
- 06 11月, 2018 4 次提交
-
-
由 hequn8128 提交于
This closes #6888.
-
由 Chesnay Schepler 提交于
-
由 yanghua 提交于
-
由 Nico Kruber 提交于
[FLINK-10655][rpc] fix RemoteRpcInvocation not overwriting ObjectInputStream's ClassNotFoundException
-
- 05 11月, 2018 12 次提交
-
-
由 Timo Walther 提交于
This closes #6857.
-
由 Timo Walther 提交于
-
由 Timo Walther 提交于
-
由 Timo Walther 提交于
-
由 Stefan Richter 提交于
This closes #7021.
-
由 Andrey Zagrebin 提交于
This closes #6957.
-
由 Piotr Nowojski 提交于
Previously there was a strict fixed order of applying LogicalCorrelateToTemporalTableJoinRule and TableScanRule rules. This was causing problems, since either of them could create a new RelNodes that have to be subject of the other rule (imagine deeply nested TemporalTableFunction that references registered tables/views and other TemporalTableFunctions). Solution to this problem is to run both of those rules in one group/collection in HepPlaner. Instead of applying one rule to whole tree then the other rule, both rules are applied to a parent node, before going down/deeper.
-
由 Stefan Richter 提交于
This closes #6975.
-
由 Till Rohrmann 提交于
Ignore the 'Artificial Failure' exceptions and rename ExceptionThrowingFailureMapper into FailureMapper to avoid false positive exception matchings.
-
由 Till Rohrmann 提交于
Remove the unnecessary call to change_scala_version.sh and remove the -Dmaven.test.skip=true property. The latter is necessary because this property suppresses the compilation and packaging of test classes. It, however, does not suppress the resolution of test dependencies which will then fail to compile because test dependencies have not been built. This commit also removes the redundant call to build flink-shaded/hadoop/flink-shaded-hadoop2-uber which is a dependency of flink-dist anyway.
-
由 Dawid Wysakowicz 提交于
* wait for whole bootstrapping script to execute on master node before submitting job * retrying to start hadoop cluster. Failling test in case could not start hadoop cluster. * added check that all containers are up and running before submitting job * reduced memory requirements for the kerberized yarn test
-
由 Stefan Richter 提交于
Please not that state written with OperatorSnapshotUtil before this commit was written in the V1 format. Now we are using the current V2. This closes #6910.
-