xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-05 00:31:23 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix v1 supported oracle for worker-cls and worker-extension-cls (#15324)

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
hijkzzz 2025-03-23 10:23:35 +08:00 committed by GitHub
parent a827aa815d
commit 0661cfef7a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 7 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
.buildkite/test-pipeline.yaml
View File

@ -148,8 +148,8 @@ steps:
# TODO: create a dedicated test section for multi-GPU example tests # TODO: create a dedicated test section for multi-GPU example tests
# when we have multiple distributed example tests # when we have multiple distributed example tests
- pushd ../examples/offline_inference - pushd ../examples/offline_inference
- python3 rlhf.py - VLLM_ENABLE_V1_MULTIPROCESSING=0 python3 rlhf.py
- RAY_DEDUP_LOGS=0 python3 rlhf_colocate.py - VLLM_ENABLE_V1_MULTIPROCESSING=0 RAY_DEDUP_LOGS=0 python3 rlhf_colocate.py
- popd - popd
- label: Metrics, Tracing Test # 10min - label: Metrics, Tracing Test # 10min

10
vllm/engine/arg_utils.py
View File

@ -1459,16 +1459,6 @@ class EngineArgs:
recommend_to_remove=False) recommend_to_remove=False)
return False return False
if self.worker_cls != EngineArgs.worker_cls:
_raise_or_fallback(feature_name="--worker-cls",
recommend_to_remove=False)
return False
if self.worker_extension_cls != EngineArgs.worker_extension_cls:
_raise_or_fallback(feature_name="--worker-extension-cls",
recommend_to_remove=False)
return False
if self.num_scheduler_steps != EngineArgs.num_scheduler_steps: if self.num_scheduler_steps != EngineArgs.num_scheduler_steps:
_raise_or_fallback(feature_name="--num-scheduler-steps", _raise_or_fallback(feature_name="--num-scheduler-steps",
recommend_to_remove=True) recommend_to_remove=True)

5
vllm/utils.py
View File

@ -2170,6 +2170,11 @@ def _maybe_force_spawn():
if cuda_is_initialized(): if cuda_is_initialized():
reason = "CUDA is initialized" reason = "CUDA is initialized"
elif is_in_ray_actor(): elif is_in_ray_actor():
# even if we choose to spawn, we need to pass the ray address
# to the subprocess so that it knows how to connect to the ray cluster.
# env vars are inherited by subprocesses, even if we use spawn.
import ray
os.environ["RAY_ADDRESS"] = ray.get_runtime_context().gcs_address
reason = "In a Ray actor and can only be spawned" reason = "In a Ray actor and can only be spawned"
if reason is not None: if reason is not None: