[
https://issues.apache.org/jira/browse/HDDS-16135?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Siyao Meng updated HDDS-16135:
------------------------------
Attachment: HDDS-16135.001.patch
> Intermittent failure in
> TestKeyLifecycleService$Normal#testNestedFSODirectoryScanResume
> ---------------------------------------------------------------------------------------
>
> Key: HDDS-16135
> URL: https://issues.apache.org/jira/browse/HDDS-16135
> Project: Apache Ozone
> Issue Type: Sub-task
> Components: Ozone Manager, test
> Reporter: Siyao Meng
> Priority: Major
> Attachments: HDDS-16135.001.patch
>
>
> {code}
> [ERROR] Tests run: 148, Failures: 1, Errors: 1, Skipped: 6, Time elapsed:
> 99.18 s <<< FAILURE! -- in
> org.apache.hadoop.ozone.om.service.TestKeyLifecycleService$Normal
> [ERROR]
> org.apache.hadoop.ozone.om.service.TestKeyLifecycleService$Normal.testNestedFSODirectoryScanResume(BucketLayout,
> int)[2][2] -- Time elapsed: 0.336 s <<< FAILURE!
> org.opentest4j.AssertionFailedError: expected: <true> but was: <false>
> at
> org.apache.hadoop.ozone.om.service.TestKeyLifecycleService$Normal.testNestedFSODirectoryScanResume(TestKeyLifecycleService.java:797)
> [ERROR]
> org.apache.hadoop.ozone.om.service.TestKeyLifecycleService$Normal.testNestedFSODirectoryScanResume(BucketLayout,
> int)[2][3] -- Time elapsed: 5.580 s <<< ERROR!
> java.util.concurrent.TimeoutException:
> {code}
> The FSO directory-scan resume mechanism itself works, but the test races the
> shared periodic {{KeyLifecycleService}} (fires every ~300ms) against a
> one-shot fault injector. Only the first scan aborts at the intended resume
> point; the next tick resumes and overwrites the persisted
> {{OmLifecycleScanState}}, so the exact-value assertions at lines 785-808
> depend on reading inside the ~300ms window (the {{[2][2]}} {{lastScannedKey}}
> failure). {{getNumKeyIterated()}} is a service-wide cumulative counter that
> concurrent re-scans inflate, so the exact-equality wait can never settle,
> giving the {{[2][3]}} timeout. Same fragility family as HDDS-16033 (different
> method). Suggested hardening: park the scan at the resume point (blocking
> injector) for deterministic inspection, and make the metric wait monotonic
> ({{>=}}) instead of exact.
> Seen in:
> -
> https://github.com/adoroszlai/ozone-build-results/blob/master/2026/08/05/51974/integration-om/hadoop-ozone/ozone-manager/org.apache.hadoop.ozone.om.service.TestKeyLifecycleService$Normal.txt
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
