[
https://issues.apache.org/jira/browse/FLINK-40790?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18119235#comment-18119235
]
sepuri sai krishna commented on FLINK-40790:
--------------------------------------------
The test assumes both `increaseResourceRequirementsBy` calls land inside
`slotRequestMaxInterval`, which is 50 ms in this parameter. Each call schedules
the declaration and cancels the pending one, so a stall between them declares
the first increment on its own. Sleeping 150 ms between the two calls
reproduces the failure exactly.
I have a fix that drives the batching window with a
`ManuallyTriggeredScheduledExecutor` instead of wall-clock time. Could you
assign this to me? PR coming shortly.
> DefaultDeclarativeSlotPoolTest.testIncreasingResourceRequirementsWillSendResourceRequirementNotification
> receives the first increment alone
> -------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-40790
> URL: https://issues.apache.org/jira/browse/FLINK-40790
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 2.4.0
> Reporter: Martijn Visser
> Priority: Major
>
> https://github.com/apache/flink/actions/runs/35493621721/job/106033628963
> (master push 2026-09-20, JDK 17, module core)
> {code}
> [ERROR]
> org.apache.flink.runtime.jobmaster.slotpool.DefaultDeclarativeSlotPoolTest.testIncreasingResourceRequirementsWillSendResourceRequirementNotification
> -- Time elapsed: 0.333 s <<< FAILURE!
> org.opentest4j.AssertionFailedError:
> expected:
> [ResourceRequirement{resourceProfile=ResourceProfile{cpuCores=1.7,
> taskHeapMemory=0 bytes, taskOffHeapMemory=0 bytes, managedMemory=0 bytes,
> networkMemory=0 bytes}, numberOfRequiredSlots=3},
> ResourceRequirement{resourceProfile=ResourceProfile{cpuCores=0,
> taskHeapMemory=0 bytes, taskOffHeapMemory=0 bytes, managedMemory=100.000mb
> (104857600 bytes), networkMemory=0 bytes}, numberOfRequiredSlots=1}]
> but was:
> [ResourceRequirement{resourceProfile=ResourceProfile{cpuCores=1.7,
> taskHeapMemory=0 bytes, taskOffHeapMemory=0 bytes, managedMemory=0 bytes,
> networkMemory=0 bytes}, numberOfRequiredSlots=1}]
> at
> org.apache.flink.runtime.jobmaster.slotpool.DefaultDeclarativeSlotPoolTest.testIncreasingResourceRequirementsWillSendResourceRequirementNotification(DefaultDeclarativeSlotPoolTest.java:88)
> {code}
> The merged expectation only applies to the slotRequestMaxInterval = 50 ms
> parameter, so this is that case: the two increaseResourceRequirementsBy calls
> were not batched into one notification.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
