Wang1rrr opened a new issue, #6966: URL: https://github.com/apache/rocketmq-dashboard/issues/6966
### Description When an AI run stream returns HTTP 200 with a non-SSE content type, `consumeRunStream` rejects it before acquiring a reader. The response body is never cancelled. A buffering gateway can therefore leave the underlying request/connection alive after the UI has already reported an error; retries repeat the leak. ### Reproduction 1. Mock fetch with a 200 `text/html` response whose `ReadableStream` remains open and records its `cancel()` callback. 2. Call either `openRunStream` or `attachRunStream`. 3. The call rejects with `llm.stream.unexpected_content_type`, but the cancel callback is not called. ### Expected behavior Discard the rejected response body while preserving the existing content-type error, even if body cancellation fails. ### Version `rocketmq-studio` at c34535db0819d02f68e50a856cb3d2614f0dc5dd. I plan to submit a focused fix with regression coverage for both stream entry points. -- 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]
