btlqql opened a new issue, #2178:
URL: https://github.com/apache/rocketmq-dashboard/issues/2178
### What happened?
Some OpenAI-compatible providers report stream failures inside an HTTP 200
SSE response, for example `data: {"error":{"message":"quota exceeded"}}`. The
server currently reads only `choices[0].delta.content`, treats the error event
as an empty token, and can finish the stream without surfacing the provider
failure.
### Expected behavior
Detect an explicit `error.message` SSE envelope and return a structured
gateway error that preserves the provider message. Legal tokenless events such
as usage metadata should remain ignored.
### Reproduction
Return the following from the configured provider and invoke streaming
completion:
```
data: {"error":{"message":"quota exceeded"}}
data: [DONE]
```
The current call completes with no tokens instead of reporting the quota
error.
--
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]
