[Misc] Add warning for incompatible Ray backend with ACL Graph mode (#2132)

### What this PR does / why we need it?

cherry-pick #1501 from 0.9.1-dev to main

Currently, Ray is not compatible with ACL Graph, so we need to fall back
to eager mode when using the Ray backend.

co-authored: Yizhou Liu <liu_yizhou@outlook.com>

- vLLM version: v0.10.0
- vLLM main:
2836dd73f1

Signed-off-by: 22dimensions <waitingwind@foxmail.com>

tests/e2e/singlecard/test_aclgraph.py

@@ -84,3 +84,11 @@ def test_deepseek_raises_error(monkeypatch: pytest.MonkeyPatch) -> None:
                        max_model_len=1024,
                        enforce_eager=False)
         assert "ACL Graph does not support deepseek" in str(excinfo.value)
+
+
+@pytest.mark.parametrize("model", MODELS)
+def test_ray_backend_sets_no_compilation(model: str) -> None:
+    runner = VllmRunner(model,
+                        enforce_eager=False,
+                        distributed_executor_backend="ray")
+    assert runner.model.llm_engine.vllm_config.compilation_config.level == 0

vllm_ascend/platform.py

@@ -153,6 +153,11 @@ class NPUPlatform(Platform):
                 "Torchair compilation enabled on NPU. Setting level to NO_COMPILATION"
             )
             compilation_config.level = CompilationLevel.NO_COMPILATION
+        elif parallel_config.distributed_executor_backend == "ray":
+            logger.warning(
+                "Ray distributed executor backend is not compatible with ACL Graph mode "
+                "right now. Setting level to NO_COMPILATION")
+            compilation_config.level = CompilationLevel.NO_COMPILATION
         else:
             logger.info(
                 "PIECEWISE compilation enabled on NPU. use_inductor not supported - "