apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[format] Valid para format error info (#4035)

Browse Source 
* feat(log):add_request_and_response_log

* 报错信息与OpenAI对齐
This commit is contained in:
xiaolei373
2025-09-12 19:05:17 +08:00
committed by GitHub
parent 88ea565aba
commit 9ac539471d
11 changed files with 435 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
fastdeploy/entrypoints/openai/api_server.py
View File

@@ -26,6 +26,7 @@ from multiprocessing import current_process
import uvicorn
import zmq
from fastapi import FastAPI, HTTPException, Request
from fastapi.exceptions import RequestValidationError
from fastapi.responses import JSONResponse, Response, StreamingResponse
from prometheus_client import CONTENT_TYPE_LATEST
@@ -40,6 +41,7 @@ from fastdeploy.entrypoints.openai.protocol import (
CompletionRequest,
CompletionResponse,
ControlSchedulerRequest,
ErrorInfo,
ErrorResponse,
ModelList,
)
@@ -56,6 +58,7 @@ from fastdeploy.metrics.metrics import (
)
from fastdeploy.metrics.trace_util import fd_start_span, inject_to_metadata, instrument
from fastdeploy.utils import (
ExceptionHandler,
FlexibleArgumentParser,
StatefulSemaphore,
api_server_logger,
@@ -232,6 +235,8 @@ async def lifespan(app: FastAPI):
app = FastAPI(lifespan=lifespan)
app.add_exception_handler(RequestValidationError, ExceptionHandler.handle_request_validation_exception)
app.add_exception_handler(Exception, ExceptionHandler.handle_exception)
instrument(app)
@@ -336,7 +341,7 @@ async def create_chat_completion(request: ChatCompletionRequest):
if isinstance(generator, ErrorResponse):
api_server_logger.debug(f"release: {connection_semaphore.status()}")
connection_semaphore.release()
return JSONResponse(content={"detail": generator.model_dump()}, status_code=generator.code)
return JSONResponse(content=generator.model_dump(), status_code=500)
elif isinstance(generator, ChatCompletionResponse):
api_server_logger.debug(f"release: {connection_semaphore.status()}")
connection_semaphore.release()
@@ -365,7 +370,7 @@ async def create_completion(request: CompletionRequest):
generator = await app.state.completion_handler.create_completion(request)
if isinstance(generator, ErrorResponse):
connection_semaphore.release()
return JSONResponse(content=generator.model_dump(), status_code=generator.code)
return JSONResponse(content=generator.model_dump(), status_code=500)
elif isinstance(generator, CompletionResponse):
connection_semaphore.release()
return JSONResponse(content=generator.model_dump())
@@ -388,7 +393,7 @@ async def list_models() -> Response:
models = await app.state.model_handler.list_models()
if isinstance(models, ErrorResponse):
return JSONResponse(content=models.model_dump(), status_code=models.code)
return JSONResponse(content=models.model_dump())
elif isinstance(models, ModelList):
return JSONResponse(content=models.model_dump())
@@ -502,7 +507,8 @@ def control_scheduler(request: ControlSchedulerRequest):
"""
Control the scheduler behavior with the given parameters.
"""
content = ErrorResponse(object="", message="Scheduler updated successfully", code=0)
content = ErrorResponse(error=ErrorInfo(message="Scheduler updated successfully", code=0))
global llm_engine
if llm_engine is None: