@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / b95697d731

Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 14, 05:30 UTC

[Frontend] improve error logging of chat completion (#22957)

Signed-off-by: Chen Zhang <zhangch99@outlook.com>

Chen Zhang committed Aug 20, 2025 at 13:03 UTC b95697d7310637399998ebf1f21a26b523aa6611

1 file changed +57 -17

vllm/entrypoints/openai/api_server.py

+57 -17

@@ -600,8 +600,11 @@ async def create_responses(request: ResponsesRequest, raw_request: Request):

600 if handler is None:

601 return base(raw_request).create_error_response(

602 message="The model does not support Responses API")

603 -

604 - generator = await handler.create_responses(request, raw_request)

603 + try:

604 + generator = await handler.create_responses(request, raw_request)

605 + except Exception as e:

606 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

607 + detail=str(e)) from e

608

609 if isinstance(generator, ErrorResponse):

610 return JSONResponse(content=generator.model_dump(),

@@ -618,7 +621,11 @@ async def retrieve_responses(response_id: str, raw_request: Request):

621 return base(raw_request).create_error_response(

622 message="The model does not support Responses API")

623

621 - response = await handler.retrieve_responses(response_id)

624 + try:

625 + response = await handler.retrieve_responses(response_id)

626 + except Exception as e:

627 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

628 + detail=str(e)) from e

629

630 if isinstance(response, ErrorResponse):

631 return JSONResponse(content=response.model_dump(),

@@ -633,7 +640,11 @@ async def cancel_responses(response_id: str, raw_request: Request):

640 return base(raw_request).create_error_response(

641 message="The model does not support Responses API")

642

636 - response = await handler.cancel_responses(response_id)

643 + try:

644 + response = await handler.cancel_responses(response_id)

645 + except Exception as e:

646 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

647 + detail=str(e)) from e

648

649 if isinstance(response, ErrorResponse):

650 return JSONResponse(content=response.model_dump(),

@@ -667,9 +678,11 @@ async def create_chat_completion(request: ChatCompletionRequest,

678 if handler is None:

679 return base(raw_request).create_error_response(

680 message="The model does not support Chat Completions API")

670 -

671 - generator = await handler.create_chat_completion(request, raw_request)

672 -

681 + try:

682 + generator = await handler.create_chat_completion(request, raw_request)

683 + except Exception as e:

684 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

685 + detail=str(e)) from e

686 if isinstance(generator, ErrorResponse):

687 return JSONResponse(content=generator.model_dump(),

688 status_code=generator.error.code)

@@ -742,7 +755,11 @@ async def create_embedding(request: EmbeddingRequest, raw_request: Request):

755 return base(raw_request).create_error_response(

756 message="The model does not support Embeddings API")

757

745 - generator = await handler.create_embedding(request, raw_request)

758 + try:

759 + generator = await handler.create_embedding(request, raw_request)

760 + except Exception as e:

761 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

762 + detail=str(e)) from e

763

764 if isinstance(generator, ErrorResponse):

765 return JSONResponse(content=generator.model_dump(),

@@ -770,8 +787,11 @@ async def create_pooling(request: PoolingRequest, raw_request: Request):

787 if handler is None:

788 return base(raw_request).create_error_response(

789 message="The model does not support Pooling API")

773 -

774 - generator = await handler.create_pooling(request, raw_request)

790 + try:

791 + generator = await handler.create_pooling(request, raw_request)

792 + except Exception as e:

793 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

794 + detail=str(e)) from e

795 if isinstance(generator, ErrorResponse):

796 return JSONResponse(content=generator.model_dump(),

797 status_code=generator.error.code)

@@ -791,7 +811,11 @@ async def create_classify(request: ClassificationRequest,

811 return base(raw_request).create_error_response(

812 message="The model does not support Classification API")

813

794 - generator = await handler.create_classify(request, raw_request)

814 + try:

815 + generator = await handler.create_classify(request, raw_request)

816 + except Exception as e:

817 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

818 + detail=str(e)) from e

819 if isinstance(generator, ErrorResponse):

820 return JSONResponse(content=generator.model_dump(),

821 status_code=generator.error.code)

@@ -820,7 +844,11 @@ async def create_score(request: ScoreRequest, raw_request: Request):

844 return base(raw_request).create_error_response(

845 message="The model does not support Score API")

846

823 - generator = await handler.create_score(request, raw_request)

847 + try:

848 + generator = await handler.create_score(request, raw_request)

849 + except Exception as e:

850 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

851 + detail=str(e)) from e

852 if isinstance(generator, ErrorResponse):

853 return JSONResponse(content=generator.model_dump(),

854 status_code=generator.error.code)

@@ -878,8 +906,12 @@ async def create_transcriptions(raw_request: Request,

906 message="The model does not support Transcriptions API")

907

908 audio_data = await request.file.read()

881 - generator = await handler.create_transcription(audio_data, request,

882 - raw_request)

909 + try:

910 + generator = await handler.create_transcription(audio_data, request,

911 + raw_request)

912 + except Exception as e:

913 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

914 + detail=str(e)) from e

915

916 if isinstance(generator, ErrorResponse):

917 return JSONResponse(content=generator.model_dump(),

@@ -919,8 +951,12 @@ async def create_translations(request: Annotated[TranslationRequest,

951 message="The model does not support Translations API")

952

953 audio_data = await request.file.read()

922 - generator = await handler.create_translation(audio_data, request,

923 - raw_request)

954 + try:

955 + generator = await handler.create_translation(audio_data, request,

956 + raw_request)

957 + except Exception as e:

958 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

959 + detail=str(e)) from e

960

961 if isinstance(generator, ErrorResponse):

962 return JSONResponse(content=generator.model_dump(),

@@ -949,7 +985,11 @@ async def do_rerank(request: RerankRequest, raw_request: Request):

985 if handler is None:

986 return base(raw_request).create_error_response(

987 message="The model does not support Rerank (Score) API")

952 - generator = await handler.do_rerank(request, raw_request)

988 + try:

989 + generator = await handler.do_rerank(request, raw_request)

990 + except Exception as e:

991 + raise HTTPException(status_code=HTTPStatus.INTERNAL_SERVER_ERROR.value,

992 + detail=str(e)) from e

993 if isinstance(generator, ErrorResponse):

994 return JSONResponse(content=generator.model_dump(),

995 status_code=generator.error.code)