[bugfix] Fix critical bug when reporting for all paths where...

[bugfix] Fix critical bug when reporting for all paths where handler.create_error_response is used (#34516)
Signed-off-by: Stanislav Kirillov <stas@nebius.com>
Co-authored-by: Stanislav Kirillov <stas@nebius.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>

vllm/entrypoints/openai/chat_completion/api_router.py

@@ -57,7 +57,7 @@ async def create_chat_completion(request: ChatCompletionRequest, raw_request: Re
     try:
         generator = await handler.create_chat_completion(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(
@@ -96,7 +96,7 @@ async def render_chat_completion(request: ChatCompletionRequest, raw_request: Re
     try:
         result = await handler.render_chat_request(request)
     except Exception as e:
-        return handler.create_error_response(e)
+        result = handler.create_error_response(e)
 
     if isinstance(result, ErrorResponse):
         return JSONResponse(content=result.model_dump(), status_code=result.error.code)

vllm/entrypoints/openai/completion/api_router.py

@@ -57,7 +57,7 @@ async def create_completion(request: CompletionRequest, raw_request: Request):
     try:
         generator = await handler.create_completion(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(
@@ -94,7 +94,7 @@ async def render_completion(request: CompletionRequest, raw_request: Request):
     try:
         result = await handler.render_completion_request(request)
     except Exception as e:
-        return handler.create_error_response(e)
+        result = handler.create_error_response(e)
 
     if isinstance(result, ErrorResponse):
         return JSONResponse(content=result.model_dump(), status_code=result.error.code)

vllm/entrypoints/openai/responses/api_router.py

@@ -66,7 +66,7 @@ async def create_responses(request: ResponsesRequest, raw_request: Request):
     try:
         generator = await handler.create_responses(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(
@@ -102,7 +102,7 @@ async def retrieve_responses(
             stream=stream,
         )
     except Exception as e:
-        return handler.create_error_response(e)
+        response = handler.create_error_response(e)
 
     if isinstance(response, ErrorResponse):
         return JSONResponse(
@@ -128,7 +128,7 @@ async def cancel_responses(response_id: str, raw_request: Request):
     try:
         response = await handler.cancel_responses(response_id)
     except Exception as e:
-        return handler.create_error_response(e)
+        response = handler.create_error_response(e)
 
     if isinstance(response, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/pooling/classify/api_router.py

@@ -35,7 +35,7 @@ async def create_classify(request: ClassificationRequest, raw_request: Request):
     try:
         generator = await handler.create_classify(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/pooling/embed/api_router.py

@@ -64,7 +64,7 @@ async def create_embedding(
     try:
         generator = await handler.create_embedding(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/pooling/pooling/api_router.py

@@ -44,7 +44,7 @@ async def create_pooling(request: PoolingRequest, raw_request: Request):
     try:
         generator = await handler.create_pooling(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/pooling/score/api_router.py

@@ -52,7 +52,7 @@ async def create_score(request: ScoreRequest, raw_request: Request):
     try:
         generator = await handler.create_score(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(
@@ -103,7 +103,7 @@ async def do_rerank(request: RerankRequest, raw_request: Request):
     try:
         generator = await handler.do_rerank(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/serve/disagg/api_router.py

@@ -67,7 +67,7 @@ async def generate(request: GenerateRequest, raw_request: Request):
     try:
         generator = await handler.serve_tokens(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(

vllm/entrypoints/serve/tokenize/api_router.py

@@ -52,7 +52,7 @@ async def tokenize(request: TokenizeRequest, raw_request: Request):
     try:
         generator = await handler.create_tokenize(request, raw_request)
     except Exception as e:
-        return handler.create_error_response(e)
+        generator = handler.create_error_response(e)
 
     if isinstance(generator, ErrorResponse):
         return JSONResponse(