[
https://issues.apache.org/jira/browse/FLINK-34718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17828019#comment-17828019
]
Wencong Liu commented on FLINK-34718:
-------------------------------------
The newly introduced DataStream operators are designed based on the mechanism
of FLIP-331, which means that the ResultPartitionType for specific operators in
a streaming job can be BLOCKING. However, the AdaptiveScheduler mandates that
the ResultPartitionType for all operators must be PIPELINED, therefore, these
operators are not suitable to be executed under the configuration of the
AdaptiveScheduler. The default scheduler for IT tests is the
{_}DefaultScheduler{_}, and I'm curious as to why it would change to the
AdaptiveScheduler. 🤔 [~rskraba]Â
> KeyedPartitionWindowedStream and NonPartitionWindowedStream
> IllegalStateException in AZP
> ----------------------------------------------------------------------------------------
>
> Key: FLINK-34718
> URL: https://issues.apache.org/jira/browse/FLINK-34718
> Project: Flink
> Issue Type: Bug
> Components: API / DataStream
> Affects Versions: 1.20.0
> Reporter: Ryan Skraba
> Priority: Critical
> Labels: test-stability
>
> [https://dev.azure.com/apache-flink/apache-flink/_build/results?buildId=58320&view=logs&j=8fd9202e-fd17-5b26-353c-ac1ff76c8f28&t=ea7cf968-e585-52cb-e0fc-f48de023a7ca&l=9646]
> 18 of the KeyedPartitionWindowedStreamITCase and
> NonKeyedPartitionWindowedStreamITCase unit tests introduced in FLINK-34543
> are failing in the adaptive scheduler profile, with errors similar to:
> {code:java}
> Mar 15 01:54:12 Caused by: java.lang.IllegalStateException: The adaptive
> scheduler supports pipelined data exchanges (violated by MapPartition
> (org.apache.flink.streaming.runtime.tasks.OneInputStreamTask) ->
> ddb598ad156ed281023ba4eebbe487e3).
> Mar 15 01:54:12 at
> org.apache.flink.util.Preconditions.checkState(Preconditions.java:215)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveScheduler.assertPreconditions(AdaptiveScheduler.java:438)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveScheduler.<init>(AdaptiveScheduler.java:356)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveSchedulerFactory.createInstance(AdaptiveSchedulerFactory.java:124)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.jobmaster.DefaultSlotPoolServiceSchedulerFactory.createScheduler(DefaultSlotPoolServiceSchedulerFactory.java:121)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.jobmaster.JobMaster.createScheduler(JobMaster.java:384)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.jobmaster.JobMaster.<init>(JobMaster.java:361)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.jobmaster.factories.DefaultJobMasterServiceFactory.internalCreateJobMasterService(DefaultJobMasterServiceFactory.java:128)
> Mar 15 01:54:12 at
> org.apache.flink.runtime.jobmaster.factories.DefaultJobMasterServiceFactory.lambda$createJobMasterService$0(DefaultJobMasterServiceFactory.java:100)
> Mar 15 01:54:12 at
> org.apache.flink.util.function.FunctionUtils.lambda$uncheckedSupplier$4(FunctionUtils.java:112)
> Mar 15 01:54:12 at
> java.util.concurrent.CompletableFuture$AsyncSupply.run(CompletableFuture.java:1604)
> Mar 15 01:54:12 ... 4 more
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
