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 8, 01:15 UTC
[ROCm][CI] Loosen RemoteOpenAIServer Startup Timeout (#34922)
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
Micah Williamson committed
Feb 19, 2026 at 23:37 UTC
f5432e35a3a4f0bd6e7d49c51a35a0a01bc32452
2 files changed
+2
-2
tests/entrypoints/openai/test_serving_chat.py
+1
-1
@@ -126,7 +126,7 @@ def gptoss_speculative_server(default_server_args: list[str]):
126
if is_aiter_found_and_supported():
127
env_dict = {"VLLM_ROCM_USE_AITER": "1"}
128
with RemoteOpenAIServer(
129
- GPT_OSS_MODEL_NAME, server_args, env_dict=env_dict
129
+ GPT_OSS_MODEL_NAME, server_args, env_dict=env_dict, max_wait_seconds=480
130
) as remote_server:
131
yield remote_server
132
tests/utils.py
+1
-1
@@ -190,7 +190,7 @@ class RemoteOpenAIServer:
190
model_loader.download_model(model_config)
191
192
self._start_server(model, vllm_serve_args, env_dict)
193
- max_wait_seconds = max_wait_seconds or 240
193
+ max_wait_seconds = max_wait_seconds or 360
194
self._wait_for_server(url=self.url_for("health"), timeout=max_wait_seconds)
195
196
def __enter__(self):