Skip to content

[BugFix] Fix step_shm_value in PD disaggregated deployment #4780

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

Merged
merged 2 commits into from
Nov 4, 2025
Merged

[BugFix] Fix step_shm_value in PD disaggregated deployment #4780

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cbecd95
[fix] fix v0 pd, let worker step_shm_value create=False
liyonghua0910 Nov 3, 2025
20461ca
Merge branch 'develop' into develop+fix_v0_pd_1103
liyonghua0910 Nov 4, 2025
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
15 changes: 3 additions & 12 deletions fastdeploy/worker/worker_process.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,12 +51,7 @@
from fastdeploy.eplb.experts_manager import RedundantExpertManager
from fastdeploy.eplb.utils import RearrangeExpertState
from fastdeploy.inter_communicator import EngineWorkerQueue as TaskQueue
from fastdeploy.inter_communicator import (
ExistTaskStatus,
IPCSignal,
ModelWeightsStatus,
shared_memory_exists,
)
from fastdeploy.inter_communicator import ExistTaskStatus, IPCSignal, ModelWeightsStatus
from fastdeploy.model_executor.layers.quantization import parse_quant_config
from fastdeploy.model_executor.utils import v1_loader_support
from fastdeploy.platforms import current_platform
Expand Down Expand Up @@ -544,16 +539,12 @@ def initialize_kv_cache(self) -> None:
def graph_optimize_and_warm_up_model(self) -> None:
self.worker.graph_optimize_and_warm_up_model()
# reset cache_messager prefilled_step signal
if self.scheduler_config.splitwise_role == "prefill":
if not envs.ENABLE_V1_KVCACHE_SCHEDULER and self.scheduler_config.splitwise_role == "prefill":
gpu_id = self.worker.model_runner.device_id
prefilled_step_name = f"splitwise_complete_prefilled_step_{self.local_rank}"
prefilled_step_idx_data = np.zeros(shape=[1], dtype=np.int32)
step_shm_value = IPCSignal(
name=prefilled_step_name,
array=prefilled_step_idx_data,
dtype=np.int32,
suffix=gpu_id,
create=not shared_memory_exists(prefilled_step_name),
name=prefilled_step_name, array=prefilled_step_idx_data, dtype=np.int32, suffix=gpu_id, create=False
)
step_shm_value.value[0] = -1

Expand Down
Loading