[
https://issues.apache.org/jira/browse/HDDS-16170?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated HDDS-16170:
----------------------------------
Labels: pull-request-available (was: )
> Intermittent failure in TestSnapshotDiffManager#testLoadJobsOnStartUp (DONE
> vs IN_PROGRESS)
> -------------------------------------------------------------------------------------------
>
> Key: HDDS-16170
> URL: https://issues.apache.org/jira/browse/HDDS-16170
> Project: Apache Ozone
> Issue Type: Sub-task
> Components: Snapshot, test
> Reporter: Siyao Meng
> Assignee: Siyao Meng
> Priority: Major
> Labels: pull-request-available
>
> {code}
> [ERROR] Tests run: 60, Failures: 1, Errors: 0, Skipped: 0 <<< FAILURE! -- in
> org.apache.hadoop.ozone.om.snapshot.TestSnapshotDiffManager
> org.apache.hadoop.ozone.om.snapshot.TestSnapshotDiffManager.testLoadJobsOnStartUp
> -- Time elapsed: 0.080 s <<< FAILURE!
> org.opentest4j.AssertionFailedError: expected: <DONE> but was: <IN_PROGRESS>
> at
> org.apache.hadoop.ozone.om.snapshot.TestSnapshotDiffManager.testLoadJobsOnStartUp(TestSnapshotDiffManager.java:1150)
> {code}
> h3. Root cause
> The test submits diff jobs, then stubs {{generateSnapshotDiffReport}} with a
> {{doAnswer}} that writes {{DONE}} to the DB inside the answer body, calls
> {{loadJobsOnStartUp()}}, and waits with {{attempt(() -> verify(spy,
> atLeast(1)).generateSnapshotDiffReport(...))}} before reading the job back
> from the DB and asserting {{DONE}}.
> Mockito records an invocation before it runs the stubbed answer. So
> {{verify(spy, atLeast(1))}} can pass as soon as the worker thread enters
> {{generateSnapshotDiffReport}}, while the answer is still executing and has
> not yet persisted {{DONE}}. The main thread then reads the job and observes
> the original {{IN_PROGRESS}}, so {{assertEquals(DONE, ...)}} fails
> intermittently.
> This is a recurrence of the symptom HDDS-9434 tried to address: that fix only
> replaced the fixed sleep with a wait on {{verify(...)}}, which still does not
> wait for the answer's DB write to complete.
> h3. Fix
> Wait on the actual observable end state the test asserts: poll
> {{getSnapshotDiffJobFromDb(...).getStatus()}} until it is {{DONE}} instead of
> waiting on the Mockito invocation. This is deterministic by construction,
> since the read that the assertion depends on is exactly what is waited on.
> - https://github.com/apache/ozone/actions/runs/31673201699/job/94364422373
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
