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 16, 09:30 UTC
[CI/Build] Simplify exception trace in api server tests (#9787)
Signed-off-by: youkaichao <youkaichao@gmail.com> Co-authored-by: youkaichao <youkaichao@gmail.com>
Yongzao committed
Oct 31, 2024 at 05:52 UTC
00d91c8a2cf3ebaf0f3ea69312f6e3882ed9f372
1 file changed
+7
-3
tests/utils.py
+7
-3
@@ -133,15 +133,19 @@ class RemoteOpenAIServer:
133
try:
134
if requests.get(url).status_code == 200:
135
break
136
- except Exception as err:
136
+ except Exception:
137
+ # this exception can only be raised by requests.get,
138
+ # which means the server is not ready yet.
139
+ # the stack trace is not useful, so we suppress it
140
+ # by using `raise from None`.
141
result = self.proc.poll()
142
if result is not None and result != 0:
139
- raise RuntimeError("Server exited unexpectedly.") from err
143
+ raise RuntimeError("Server exited unexpectedly.") from None
144
145
time.sleep(0.5)
146
if time.time() - start > timeout:
147
raise RuntimeError(
144
- "Server failed to start in time.") from err
148
+ "Server failed to start in time.") from None
149
150
@property
151
def url_root(self) -> str: