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

[V1] Add RayExecutor support for AsyncLLM (api server) #11712

Merged
merged 1 commit into from
Jan 4, 2025
Merged

[V1] Add RayExecutor support for AsyncLLM (api server) #11712

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
7 changes: 6 additions & 1 deletion vllm/v1/engine/async_llm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@
from vllm.v1.engine.detokenizer import Detokenizer
from vllm.v1.engine.processor import Processor
from vllm.v1.executor.abstract import Executor
from vllm.v1.executor.ray_utils import initialize_ray_cluster
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should be a lazy import right? Doesn't this import ray?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This will import Ray but won't error out if Ray is not installed:

try:
    import ray
    ...
except ImportError:
    ray = None

It errors out only when initialize_ray_cluster is called:

def initialize_ray_cluster():
    assert_ray_available()


logger = init_logger(__name__)

Expand Down Expand Up @@ -150,7 +151,11 @@ def _get_executor_cls(cls, vllm_config: VllmConfig) -> Type[Executor]:
executor_class: Type[Executor]
distributed_executor_backend = (
vllm_config.parallel_config.distributed_executor_backend)
if distributed_executor_backend == "mp":
if distributed_executor_backend == "ray":
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Instead of repeating this logic in both LLMEngine and AsyncLLM, can we instead move _get_executor_cls into abstract Executor? That way, we don't need to repeat logic across implementations. This is not a @classmethod anyways

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good catch. Yeah we should unify them. cc @ruisearch42

initialize_ray_cluster(vllm_config.parallel_config)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is called in RayExecutor constructor. No need to have it here.

from vllm.v1.executor.ray_executor import RayExecutor
executor_class = RayExecutor
elif distributed_executor_backend == "mp":
from vllm.v1.executor.multiproc_executor import MultiprocExecutor
executor_class = MultiprocExecutor
else:
Expand Down
Loading