smengcl opened a new pull request, #11443: URL: https://github.com/apache/ozone/pull/11443
Generated-by: Claude Code (Opus 5.5) ## What changes were proposed in this pull request? `TestKeyInputStream#testSkip` writes a RATIS/THREE key and reads it right away, asserting an exact ReadChunk count from `XceiverClientMetrics`. A follower may not have applied the PutBlock yet, so a ReadChunk sent to it fails with `UNKNOWN_BCSID` and the client retries on another datanode. The metric counts every attempt, so the test intermittently sees 4 ReadChunk ops instead of 2: ``` WARN keyvalue.KeyValueHandler - Operation: ReadChunk , Message: Unable to find the block with bcsID 485. Container 9 bcsId is 483. , Result: UNKNOWN_BCSID ``` The fix closes the key's container and waits until it is CLOSED on all replicas before reading (`OzoneTestHelper.waitForContainerClose`, already used by `testReadAfterReplication`). Followers apply the close only after the PutBlock (Ratis log order), so no replica can lag behind the block's bcsId. ## What is the link to the Apache JIRA https://issues.apache.org/jira/browse/HDDS-16781 ## How was this patch tested? - `mvn -pl :ozone-integration-test -am test -Dtest=TestKeyInputStream`: 4/4 passed locally. - The fix removes the race by construction; a single green run cannot prove non-flakiness on its own. - 10x10 [flaky-test-check](https://github.com/smengcl/hadoop-ozone/actions/runs/37740530899) on this branch: in progress 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
