[
https://issues.apache.org/jira/browse/FLINK-36295?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17882688#comment-17882688
]
Matthias Pohl commented on FLINK-36295:
---------------------------------------
{quote}There is a strange observation in the logs where the
StateTransitionManager is transitioning into Stabilizing phase with "Desired
resources are not met" log message indicating that there are not enough
resources available for immediate state transitioning. But the 4 task slots
become available earlier (indicated by the DefaultDeclarativeSlotPool with
"Acquired new resources; new total acquired resources:
ResourceCounter{resources={ResourceProfile {UNKNOWN}=4}}"){quote}
I came up with a reasoning for this item. I reverted FLINK-36279 locally for
the test run which means that only free slots (i.e. 2 rather than the 4 because
2 where already allocated in the Executing state) were considered for the
desired resources check. That explains why that behavior.
> AdaptiveSchedulerClusterITCase. testCheckpointStatsPersistedAcrossRescale
> failed with
> --------------------------------------------------------------------------------------
>
> Key: FLINK-36295
> URL: https://issues.apache.org/jira/browse/FLINK-36295
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Affects Versions: 2.0-preview
> Reporter: Matthias Pohl
> Assignee: Matthias Pohl
> Priority: Blocker
> Labels: test-stability
> Attachments:
> FLINK-36295.failure.62156.20240916.1.logs-cron_jdk17-test_cron_jdk17_core-1726454552.log,
> FLINK-36295.failure.with-revert.debug.log
>
>
> https://dev.azure.com/apache-flink/apache-flink/_build/results?buildId=62156&view=logs&j=675bf62c-8558-587e-2555-dcad13acefb5&t=5878eed3-cc1e-5b12-1ed0-9e7139ce0992&l=10234
> {code}
> Sep 16 03:06:30 03:06:30.168 [ERROR] Tests run: 3, Failures: 0, Errors: 1,
> Skipped: 0, Time elapsed: 5.275 s <<< FAILURE! -- in
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveSchedulerClusterITCase
> Sep 16 03:06:30 03:06:30.168 [ERROR]
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveSchedulerClusterITCase.testCheckpointStatsPersistedAcrossRescale
> -- Time elapsed: 0.676 s <<< ERROR!
> Sep 16 03:06:30 java.lang.IndexOutOfBoundsException: Index: -1
> Sep 16 03:06:30 at
> java.base/java.util.Collections$EmptyList.get(Collections.java:4586)
> Sep 16 03:06:30 at
> org.apache.flink.runtime.scheduler.adaptive.AdaptiveSchedulerClusterITCase.testCheckpointStatsPersistedAcrossRescale(AdaptiveSchedulerClusterITCase.java:214)
> Sep 16 03:06:30 at
> java.base/java.lang.reflect.Method.invoke(Method.java:568)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.RecursiveAction.exec(RecursiveAction.java:194)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.ForkJoinTask.doExec(ForkJoinTask.java:373)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(ForkJoinPool.java:1182)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.ForkJoinPool.scan(ForkJoinPool.java:1655)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.ForkJoinPool.runWorker(ForkJoinPool.java:1622)
> Sep 16 03:06:30 at
> java.base/java.util.concurrent.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:165)
> Sep 16 03:06:30
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
