Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[Core] Fix engine-use-ray broken #4105

Merged
merged 2 commits into from
Apr 16, 2024
Merged

[Core] Fix engine-use-ray broken #4105

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9be6f19
fixed a bug
rkooo567 Apr 16, 2024
a8ee188
improve testing
rkooo567 Apr 16, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 13 additions & 4 deletions tests/async_engine/test_api_server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,21 +25,30 @@ def _query_server_long(prompt: str) -> dict:


@pytest.fixture
def api_server(tokenizer_pool_size: int):
def api_server(tokenizer_pool_size: int, engine_use_ray: bool,
worker_use_ray: bool):
script_path = Path(__file__).parent.joinpath(
"api_server_async_engine.py").absolute()
uvicorn_process = subprocess.Popen([
commands = [
sys.executable, "-u",
str(script_path), "--model", "facebook/opt-125m", "--host",
"127.0.0.1", "--tokenizer-pool-size",
str(tokenizer_pool_size)
])
]
if engine_use_ray:
commands.append("--engine-use-ray")
if worker_use_ray:
commands.append("--worker-use-ray")
uvicorn_process = subprocess.Popen(commands)
yield
uvicorn_process.terminate()


@pytest.mark.parametrize("tokenizer_pool_size", [0, 2])
def test_api_server(api_server, tokenizer_pool_size: int):
@pytest.mark.parametrize("worker_use_ray", [False, True])
@pytest.mark.parametrize("engine_use_ray", [False, True])
def test_api_server(api_server, tokenizer_pool_size: int, worker_use_ray: bool,
engine_use_ray: bool):
"""
Run the API server and test it.

Expand Down
7 changes: 3 additions & 4 deletions vllm/engine/async_llm_engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -333,8 +333,7 @@ def from_engine_args(
if engine_config.device_config.device_type == "neuron":
raise NotImplementedError("Neuron is not supported for "
"async engine yet.")
elif (engine_config.parallel_config.worker_use_ray
or engine_args.engine_use_ray):
elif engine_config.parallel_config.worker_use_ray:
initialize_ray_cluster(engine_config.parallel_config)
from vllm.executor.ray_gpu_executor import RayGPUExecutorAsync
executor_class = RayGPUExecutorAsync
Expand Down Expand Up @@ -410,8 +409,8 @@ def _init_engine(self, *args,
else:
# FIXME(woosuk): This is a bit hacky. Be careful when changing the
# order of the arguments.
cache_config = args[1]
parallel_config = args[2]
cache_config = kwargs["cache_config"]
parallel_config = kwargs["parallel_config"]
if parallel_config.tensor_parallel_size == 1:
num_gpus = cache_config.gpu_memory_utilization
else:
Expand Down
Loading