[
https://issues.apache.org/jira/browse/FLINK-37701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17959556#comment-17959556
]
Aleksandr Iushmanov commented on FLINK-37701:
---------------------------------------------
I can see 2 problems breaking this test.
1. Execution graph for some reason goes through `Cancelling -> Cancelled`
states before job resubmission (which doesn't match my expectations based on
docs !screenshot-1.png! ). Going through `terminal` state it `nulls` checkpoint
coordinator, hence StateSizeEstimate class can completely ignore last
checkpoint.
2. Test job doesn't have any `keyedManagedState` and `StateSizeEstimate` scorer
gives 0. This way we score matching key group allocations same 0 as
non-matching, which leads to random slot allocation.
I have raised this PR as a discussion starter. [~roman], please let me know
what do you think?
https://github.com/apache/flink/pull/26663
> The testRecoverLocallyFromProcessCrashWithWorkingDirectory test failed of
> azure cron adaptive scheduler pipeline
> -----------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-37701
> URL: https://issues.apache.org/jira/browse/FLINK-37701
> Project: Flink
> Issue Type: Bug
> Components: Build System / Azure Pipelines, Build System / CI
> Affects Versions: 2.1.0
> Reporter: dalongliu
> Assignee: Aleksandr Iushmanov
> Priority: Major
> Labels: pull-request-available
> Fix For: 2.1.0
>
> Attachments: screenshot-1.png
>
>
> The detail:
> https://dev.azure.com/apache-flink/apache-flink/_build/results?buildId=67293&view=logs&j=8fd9202e-fd17-5b26-353c-ac1ff76c8f28&t=ea7cf968-e585-52cb-e0fc-f48de023a7ca
> {code:java}
> Apr 20 03:21:57 03:21:57.387 [ERROR] Tests run: 1, Failures: 1, Errors: 0,
> Skipped: 0, Time elapsed: 17.77 s <<< FAILURE! -- in
> org.apache.flink.test.recovery.LocalRecoveryITCase
> Apr 20 03:21:57 03:21:57.387 [ERROR]
> org.apache.flink.test.recovery.LocalRecoveryITCase.testRecoverLocallyFromProcessCrashWithWorkingDirectory
> -- Time elapsed: 17.74 s <<< FAILURE!
> Apr 20 03:21:57 org.opentest4j.AssertionFailedError: [The task was deployed
> to AllocationID(bb6371bf3fe9fbcb2ee329893e802fde) but it should have been
> deployed to AllocationID(5100f7baf1dea42453fd9b1c17d6d732) for local
> recovery., The task was deployed to
> AllocationID(e357fcd5041e52b7e647ca463cfe471a) but it should have been
> deployed to AllocationID(bb6371bf3fe9fbcb2ee329893e802fde) for local
> recovery., The task was deployed to
> AllocationID(5100f7baf1dea42453fd9b1c17d6d732) but it should have been
> deployed to AllocationID(e357fcd5041e52b7e647ca463cfe471a) for local
> recovery.] ==> expected: <true> but was: <false>
> Apr 20 03:21:57 at
> org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
> Apr 20 03:21:57 at
> org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)
> Apr 20 03:21:57 at
> org.junit.jupiter.api.AssertTrue.failNotTrue(AssertTrue.java:63)
> Apr 20 03:21:57 at
> org.junit.jupiter.api.AssertTrue.assertTrue(AssertTrue.java:36)
> Apr 20 03:21:57 at
> org.junit.jupiter.api.Assertions.assertTrue(Assertions.java:214)
> Apr 20 03:21:57 at
> org.apache.flink.test.recovery.LocalRecoveryITCase.testRecoverLocallyFromProcessCrashWithWorkingDirectory(LocalRecoveryITCase.java:119)
> Apr 20 03:21:57 at
> java.base/java.lang.reflect.Method.invoke(Method.java:568)
> Apr 20 03:21:57 at
> java.base/java.util.concurrent.ForkJoinTask.doExec(ForkJoinTask.java:373)
> Apr 20 03:21:57 at
> java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(ForkJoinPool.java:1182)
> Apr 20 03:21:57 at
> java.base/java.util.concurrent.ForkJoinPool.scan(ForkJoinPool.java:1655)
> Apr 20 03:21:57 at
> java.base/java.util.concurrent.ForkJoinPool.runWorker(ForkJoinPool.java:1622)
> Apr 20 03:21:57 at
> java.base/java.util.concurrent.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:165)
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
