上下文超长场景下，未能正确返回错误码，proxy server返回200-InferNex-AtomGit

已关闭

上下文超长场景下，未能正确返回错误码，proxy server返回200 #143

ds_13571877104创建于 8月4日关闭于 8月14日

8月4日创建

2026-08-04 12:55:37,850 - httpx - INFO - HTTP Request: POST http://x.x.x.x:8000/v1/chat/completions "HTTP/1.1 400 Bad Request"
2026-08-04 12:55:37,851 - __main__ - WARNING - Decoder stream attempt 1 failed for /v1/chat/completions: Client error '400 Bad Request' for url 'http://x.x.x.x:8000/v1/chat/completions'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
2026-08-04 12:55:38,444 - httpx - INFO - HTTP Request: POST http://x.x.x.x:8000/v1/chat/completions "HTTP/1.1 400 Bad Request"
2026-08-04 12:55:38,445 - __main__ - WARNING - Decoder stream attempt 2 failed for /v1/chat/completions: Client error '400 Bad Request' for url 'http://x.x.x.x:8000/v1/chat/completions'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
2026-08-04 12:55:39,049 - httpx - INFO - HTTP Request: POST http://x.x.x.x:8000/v1/chat/completions "HTTP/1.1 400 Bad Request"
2026-08-04 12:55:39,050 - __main__ - WARNING - Decoder stream attempt 3 failed for /v1/chat/completions: Client error '400 Bad Request' for url 'http://x.x.x.x:8000/v1/chat/completions'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400
2026-08-04 12:55:39,050 - __main__ - ERROR - Decoder failed after 3 attempts streaming to /v1/chat/completions.
2026-08-04 12:55:39,050 - __main__ - ERROR - Error streaming from decoder http://x.x.x.x:8000 for request 53a929c8-791d-4f63-acb3-2e447fed5cb6: Client error '400 Bad Request' for url 'http://x.x.x.x:8000/v1/chat/completions'
For more information check: https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400. The aborted request will be recorded for KV release.

curl -X POST http://xxxxx:8000/v1/chat/completions -H "Content-Type: application/json" -d @payload.json -vvvv
Note: Unnecessary use of -X or --request, POST is already inferred.
* Uses proxy env variable no_proxy == '127.0.0.1,localhost,*.huawei.com '
*   Trying xxxxx:8000...
* Connected to xxx (xxxx) port 8000 (#0)
> POST /v1/chat/completions HTTP/1.1
> Host: xxx:8000
> User-Agent: curl/7.79.1
> Accept: */*
> Content-Type: application/json
> Content-Length: 483708
> 
* We are completely uploaded and fine
* Mark bundle as not supporting multiuse
< HTTP/1.1 200 OK
< date: Tue, 04 Aug 2026 12:55:35 GMT
< server: uvicorn
< content-type: application/json
< Transfer-Encoding: chunked
< 
* Connection #0 to host xxx left intact