BenRobinson-hub commented on issue #10911: URL: https://github.com/apache/rocketmq/issues/10911#issuecomment-6099313937
We ran a bounded follow-up against the official RocketMQ `develop` source at commit `9ea2ccdc2b41445aee5e177e14e501d606f5a1ed`, with `rocketmq-apis` fixed at `3e60073c6dab3430feaec443824802e23ecda681`. The test used three isolated, non-production Broker runs: - **C0:** `serverNettyWorkerGroupEnable=true`, `serverPooledByteBufAllocatorEnable=true` - **C1:** worker group disabled, pooled allocator enabled - **C2:** worker group enabled, pooled allocator disabled Each run accepted and read back **451 messages**. Across the three runs, **1,353 / 1,353** sent message bodies matched the corresponding parsed CommitLog bodies by SHA-256, and the RocketMQ-style 31-bit `BODYCRC` values also matched **1,353 / 1,353**. We did not observe any candidate body containing markers attributable to the independent pull connections used by the test. So the bounded result is: **NOT_REPRODUCED_IN_THIS_BOUNDED_RUN** A few important limitations: - this was a short bounded run, not a long-duration stress test; - it used current official `develop`, not the historical customized 4.9.8 environment described in this issue; - raw on-disk CommitLog bytes were not independently preserved for a second disk-level forensic pass; - this result does **not** prove that the historical defect did not exist or that it has been fixed. We are sharing this only as an additional data point for the issue. -- 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]
