[
https://issues.apache.org/jira/browse/FLINK-39413?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18072361#comment-18072361
]
Mukul Gupta commented on FLINK-39413:
-------------------------------------
The test takes ~20s because NETWORK_PARTITION_REQUEST_TIMEOUT is not set and
defaults to 10s. The ResultPartitionManager's listener timeout checker runs
every 10s with strict greater-than expiration, so it
takes two full cycles (~20s) before the PartitionNotFoundException reaches the
client.
Fix: Set NETWORK_PARTITION_REQUEST_TIMEOUT to 100ms in the test config. The
test still validates the same retry-and-fail path but completes in under 1s.
Could you please assign this to me?
> TaskExecutorSubmissionTest#testRemotePartitionNotFound needs 20s to finish
> --------------------------------------------------------------------------
>
> Key: FLINK-39413
> URL: https://issues.apache.org/jira/browse/FLINK-39413
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Coordination
> Affects Versions: 2.0.1, 2.2.0, 2.1.1
> Reporter: Matthias Pohl
> Priority: Minor
> Labels: test-stability
>
> {{TaskExecutorSubmissionTest#testRemotePartitionNotFound}} takes 20s to
> finish. We might want to look into speeding up the test.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
