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 15, 04:45 UTC
[CI/Build][Bugfix] Fix deadlock on v1 engine test CI (#19872)
Signed-off-by: Isotr0py <2037008807@qq.com>
Isotr0py committed
Jun 20, 2025 at 09:51 UTC
ee9a1531aad9bc5cb585a230b467927bfaa5fc39
1 file changed
+2
-1
tests/v1/engine/test_async_llm.py
+2
-1
@@ -383,7 +383,8 @@ async def test_check_health(monkeypatch: pytest.MonkeyPatch):
383
with monkeypatch.context() as m, ExitStack() as after:
384
m.setenv("VLLM_USE_V1", "1")
385
386
- engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
386
+ with set_default_torch_num_threads(1):
387
+ engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
388
after.callback(engine.shutdown)
389
390
# Test 1: Healthy engine should not raise any exception