[Enhancement] Engine runtime errors (#2426)

Signed-off-by: Daniel Huang <daniel1.huang@intel.com>
Signed-off-by: Daniel Huang <pilotflyer824@gmail.com>
Signed-off-by: bjf-frz <frz123db@gmail.com>
Co-authored-by: Chenguang Zheng <645327136@qq.com>
Co-authored-by: Zhou Taichang <tzhouam@connect.ust.hk>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
Co-authored-by: bjf-frz <frz123db@gmail.com>

tests/entrypoints/openai_api/test_image_server.py

@@ -407,6 +407,29 @@ def test_health_endpoint_no_engine():
     assert data["status"] == "unhealthy"
 
 
+def test_health_endpoint_dead_engine():
+    """Health returns 503 when the engine raises EngineDeadError."""
+    from unittest.mock import AsyncMock
+
+    from fastapi import FastAPI
+    from vllm.v1.engine.exceptions import EngineDeadError
+
+    from vllm_omni.entrypoints.openai.api_server import router
+
+    app = FastAPI()
+    app.include_router(router)
+
+    dead_engine = AsyncMock()
+    dead_engine.check_health = AsyncMock(side_effect=EngineDeadError())
+    app.state.engine_client = dead_engine
+
+    client = TestClient(app)
+    response = client.get("/health")
+    assert response.status_code == 503
+    data = response.json()
+    assert data["status"] == "unhealthy"
+
+
 def test_models_endpoint(test_client):
     """Test /v1/models endpoint for diffusion mode"""
     response = test_client.get("/v1/models")

tests/entrypoints/openai_api/test_video_server.py

@@ -133,7 +133,6 @@ def _wait_for_status(client: TestClient, video_id: str, status: str, timeout_s:
     last_payload = None
     while time.time() < deadline:
         response = client.get(f"/v1/videos/{video_id}")
-        assert response.status_code == 200
         last_payload = response.json()
         if last_payload["status"] == status:
             return last_payload
@@ -644,7 +643,7 @@ def test_invalid_lora_returns_400(test_client):
     assert response.status_code == 200
     video_id = response.json()["id"]
     failed = _wait_for_status(test_client, video_id, VideoGenerationStatus.FAILED.value)
-    assert failed["error"]["code"] == "HTTPException"
+    assert failed["error"]["code"] == 400
     assert "lora object" in failed["error"]["message"].lower()