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 5, 16:00 UTC
[Bugfix] Preserve original ImportError in gRPC server entrypoint (#38673)
Signed-off-by: Chang Su <chang.s.su@oracle.com>
Chang Su committed
Apr 1, 2026 at 15:16 UTC
54500546ac2c34d650ae6d7757777a005dfbc4c0
1 file changed
+6
-4
vllm/entrypoints/grpc_server.py
+6
-4
@@ -29,11 +29,13 @@ try:
29
from grpc_reflection.v1alpha import reflection
30
from smg_grpc_proto import vllm_engine_pb2, vllm_engine_pb2_grpc
31
from smg_grpc_servicer.vllm.servicer import VllmEngineServicer
32
-except ImportError:
32
+except ImportError as e:
33
raise ImportError(
34
- "smg-grpc-servicer is required for gRPC mode. "
35
- "Install it with: pip install vllm[grpc]"
36
- ) from None
34
+ "gRPC mode requires smg-grpc-servicer. "
35
+ "If not installed, run: pip install vllm[grpc]. "
36
+ "If already installed, there may be a broken import due to a "
37
+ "version mismatch — see the chained exception above for details."
38
+ ) from e
39
40
import uvloop
41