[Feature] Failure message shows more details (#2961)

Signed-off-by: wuhang <wuhang6@huawei.com>
This commit is contained in:
wuhang authored and GitHub committed 2026-04-23 04:32:43 +08:00
1 parent e3b0afbb97
commit fd9a0ee27f
6 files changed
+205 -38

No files matched your search

+20 -3
View File
@@ -14,6 +14,7 @@ from vllm.v1.engine.exceptions import EngineDeadError, EngineGenerateError
from vllm_omni.entrypoints.async_omni import AsyncOmni
from vllm_omni.entrypoints.omni import Omni
from vllm_omni.entrypoints.omni_base import OmniEngineDeadError
pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
@@ -335,6 +336,18 @@ def _enqueue_error_message(engine: FakeAsyncOmniEngine, msg: dict[str, Any]) ->
)
def _enqueue_fatal_error_message(engine: FakeAsyncOmniEngine, msg: dict[str, Any]) -> None:
engine.output_q.put_nowait(
{
"type": "error",
"fatal": True,
"request_id": msg["request_id"],
"stage_id": 2,
"error": "engine dead",
}
)
@pytest.mark.asyncio
async def test_get_supported_tasks_returns_engine_supported_tasks():
omni = object.__new__(AsyncOmni)
@@ -546,18 +559,22 @@ async def test_async_omni_abort_forwards_to_engine(monkeypatch: pytest.MonkeyPat
@pytest.mark.asyncio
async def test_async_omni_propagates_engine_error(monkeypatch: pytest.MonkeyPatch):
engine = FakeAsyncOmniEngine(stage_metadata=THREE_STAGE_META, on_add_request=_enqueue_error_message)
async def test_async_omni_propagates_fatal_error_context(monkeypatch: pytest.MonkeyPatch):
engine = FakeAsyncOmniEngine(stage_metadata=THREE_STAGE_META, on_add_request=_enqueue_fatal_error_message)
_patch_engine(monkeypatch, engine)
app = AsyncOmni("dummy-model")
try:
with pytest.raises(RuntimeError, match="engine boom"):
with pytest.raises(EngineDeadError, match="engine dead") as exc_info:
async for _ in app.generate(prompt="hello", request_id="req-1"):
pass
finally:
app.shutdown()
assert isinstance(exc_info.value, OmniEngineDeadError)
assert str(exc_info.value) == "engine dead"
assert getattr(exc_info.value, "error_stage_id") == 2
def test_omni_generate_py_generator_yields_final_outputs_for_each_request(monkeypatch: pytest.MonkeyPatch):
sampling_params = [SamplingParams(max_tokens=8) for _ in range(3)]