DanielLeens commented on code in PR #11557: URL: https://github.com/apache/seatunnel/pull/11557#discussion_r4219672511
########## docs/zh/engines/zeta/managed-source-runtime.md: ########## @@ -0,0 +1,206 @@ +--- +sidebar_position: 16 +--- + +# 引擎受管 Source 运行时 + +<!-- +Licensed to the Apache Software Foundation (ASF) under one or more +contributor license agreements. See the NOTICE file distributed with +this work for additional information regarding copyright ownership. +The ASF licenses this file to You under the Apache License, Version 2.0 +(the "License"); you may not use this file except in compliance with +the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. +--> + +引擎受管 Source 运行时是 [STIP-31](https://github.com/apache/seatunnel/issues/11558) +引入的实验性 Zeta 执行通道。它把 Reader 和 Split Enumerator 的 checkpoint 可见状态统一放入 +引擎拥有的事件循环中串行执行，使连接器不再需要自行用 checkpoint lock 协调引擎回调。 + +该功能默认关闭。只有以下条件同时满足，现有作业才会在新部署时进入受管通道： + +1. `managed-source-runtime.enabled` 为 `true`。 +2. 连接器名称在 `connector-allowlist` 中。 Review Comment: Thanks @nzw921rx for the review. On the switch: the current design is a default-off, experimental gate with two parts, `managed-source-runtime.enabled` and a per-connector `connector-allowlist`, so that an unsupported connector never silently enters the managed path. Your concern is fair: exposing the allowlist lets a user opt an unsupported connector in, and as connectors migrate the switch loses meaning. Removing it would move the "does this connector support the managed runtime" decision into the connector contract itself (fail fast at job start), which is a design change to STIP-31 rather than a local edit. I'd rather not change that unilaterally in this PR, so I'll raise it on the STIP-31 design issue for the maintainers to weigh in. If the consensus is to drop the switch, I'll update the code and the docs in both languages. On the benchmark: there is no benchmark in this PR today, and I have not measured the event-loop overhead. I'll note it as a follow-up item rather than claim any performance numbers here. -- 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]
