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 9, 16:00 UTC
[Perf] Cache exc.errors() result in validation exception handler (#32984)
Signed-off-by: 7. Sun <jhao.sun@gmail.com>
7. Sun committed
Jan 24, 2026 at 10:01 UTC
0f19427db587ae425d62a1857ee3339015ee3084
1 file changed
+4
-3
vllm/entrypoints/openai/api_server.py
+4
-3
@@ -558,7 +558,8 @@ def build_app(args: Namespace) -> FastAPI:
558
@app.exception_handler(RequestValidationError)
559
async def validation_exception_handler(_: Request, exc: RequestValidationError):
560
param = None
561
- for error in exc.errors():
561
+ errors = exc.errors()
562
+ for error in errors:
563
if "ctx" in error and "error" in error["ctx"]:
564
ctx_error = error["ctx"]["error"]
565
if isinstance(ctx_error, VLLMValidationError):
@@ -566,9 +567,9 @@ def build_app(args: Namespace) -> FastAPI:
567
break
568
569
exc_str = str(exc)
569
- errors_str = str(exc.errors())
570
+ errors_str = str(errors)
571
571
- if exc.errors() and errors_str and errors_str != exc_str:
572
+ if errors and errors_str and errors_str != exc_str:
573
message = f"{exc_str} {errors_str}"
574
else:
575
message = exc_str