mwkang commented on PR #8469: URL: https://github.com/apache/hbase/pull/8469#issuecomment-5112896500
Thanks for catching this. The failure was legitimate. The test held the old entry's offset write lock while waiting for the replacement block to be fully flushed. After `c72699b800`, the writer also needs that lock to finish `putIntoBackingMap`, so both sides waited on each other. I pushed `35db4868fe` to fix the test's lock ordering. The test now waits only for the replacement to be published while holding the old entry's lock, releases the lock, and then waits for the full flush. It still verifies that the pending eviction does not remove the replacement entry and that `blocksByHFile` remains consistent. I ran the full `TestBucketCache` suite locally with JDK 17: 68 tests, 0 failures, and 0 errors. -- 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]
