wang-jiahua commented on PR #10514: URL: https://github.com/apache/rocketmq/pull/10514#issuecomment-5276027880
## Re-validated on a Fresh Environment (2026-08-13) Independent re-run on newly provisioned hardware, same methodology as #10444. **Environment**: `ecs.g7.2xlarge` (8C32G) broker with a 500GB ESSD PL1 store mounted `data=writeback`, separate producer host, Dragonwell 21.0.11, 1KB messages, sync send, 256 threads, store wiped and broker restarted per run. Baseline is develop `e3458616d`. ### Latency, 90s, ~13M messages, 0 failures | Build | TPS | P50 | P90 | P99 | P999 | Max | Broker Young GC | |---|---|---|---|---|---|---|---| | develop | 141724 | 1.57 | 2.54 | 5.86 | 12.96 | 351.93 | 88 | | this PR | 142036 | 1.56 | 2.60 | 5.79 | 12.45 | 359.09 | 82 | Milliseconds. Broker Young GC is down 6.8%, P99 down 1.2%, P999 down 3.9%, TPS up 0.2%. An earlier throughput-only run on the same setup showed 169670 TPS against 159678 for develop, a 6.3% gap, but I would not lean on that number: repeat runs here move within roughly +/-6%, so a single 6% delta sits at the edge of the noise band rather than clearly outside it. The GC count is the steadier of the two signals and moves in the same direction across both runs. No send failures in any run. -- 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]
