sanity test

Signed-off-by: Erin Ho <[email protected]>

precheck

Author: hchings

tensorrt_llm/executor/rpc/rpc_client.py

@@ -412,7 +412,8 @@ def custom_exception_handler(loop, context):
                 exception = context.get('exception')
                 message = context.get('message', '')
 
-                if isinstance(exception, asyncio.CancelledError) or "pending" in message:
+                if isinstance(exception,
+                              asyncio.CancelledError) or "pending" in message:
                     logger.debug(f"Suppressed error during shutdown: {message}")
                     return
 

tests/integration/defs/examples/test_ray.py

@@ -12,7 +12,11 @@ def ray_example_root(llm_root):
     return example_root
 
 
-def test_llm_inference_async_ray(ray_example_root, llm_venv):
+@pytest.mark.parametrize("use_rpc", [True, False], ids=["rpc", "no_rpc"])
+def test_llm_inference_async_ray(ray_example_root, llm_venv, monkeypatch,
+                                 use_rpc):
+    if use_rpc:
+        monkeypatch.setenv("TLLM_RAY_USE_RPC", "1")
     script_path = os.path.join(ray_example_root, "llm_inference_async_ray.py")
     model_path = f"{llm_models_root()}/llama-models-v2/TinyLlama-1.1B-Chat-v1.0"
     venv_check_call(llm_venv, [script_path, "--model", model_path])