Aias00 opened a new pull request, #7023: URL: https://github.com/apache/rocketmq-dashboard/pull/7023
### Which Issue(s) This PR Fixes Fixes #7017 ### Brief Description Use a production emitter factory that moves socket writes off run and heartbeat/flush threads. Each observer has one ordered application queue bounded to 256 frames / 1 MiB estimated text memory, plus an in-flight frame. The sender pool is capped at 128 daemon threads with no executor waiting queue and no caller-runs fallback. Overflow/failure closes only that observer; completion drains admitted frames in order. Timeout/disconnect clears pending frames; server shutdown interrupts senders. Existing replay/dedup and run cancellation code is unchanged. ### How Did You Test This Change? - The synchronous transport failed the blocked-observer and budget regressions; no isolation was inferred from mocks of the registry itself. - Java21 conversation suite: 215 tests passed; 0 Checkstyle violations. - Six new transport tests exercise real emitters and registry/session fan-out with controlled handler writes: slow/healthy observer isolation, FIFO + done/complete-once, frame/byte overflow, timeout cleanup, shutdown and rejected sender execution. - git diff --check passed; all test sources compiled. No real-network load benchmark or full backend suite was run locally. Blocked container writes still depend on servlet/socket timeout behavior; interruption is not promised to release every OS write immediately. The budget covers the application queue, not servlet/proxy internal buffers. Architecture documentation records these boundaries. No wire changes or dependencies. ### Checklist - [x] One coherent change - [x] Conventional Commit; ...Test regression names - [x] New Java files have ASF headers - [x] Resource limits and shutdown behavior documented -- 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]
