From 12001f2ebc606b471476d47edc22a79af6aca66c Mon Sep 17 00:00:00 2001 From: maobaolong Date: Thu, 12 Mar 2026 04:45:20 +0800 Subject: [PATCH] [LMCache] Pass TP size in lookup for MLA multi-reader locking (#36129) Signed-off-by: baoloongmao Co-authored-by: Yihua Cheng --- .../lmcache_integration/multi_process_adapter.py | 6 ++++++ .../kv_connector/v1/lmcache_mp_connector.py | 16 ++++++++++++++++ 2 files changed, 22 insertions(+) diff --git a/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_integration/multi_process_adapter.py b/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_integration/multi_process_adapter.py index e476cba7cd..eff580df90 100644 --- a/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_integration/multi_process_adapter.py +++ b/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_integration/multi_process_adapter.py @@ -114,6 +114,7 @@ class LMCacheMPSchedulerAdapter: world_size: int, kv_rank: int, vllm_block_size: int, + tp_size: int = 1, ): """ Args: @@ -124,6 +125,8 @@ class LMCacheMPSchedulerAdapter: world_size: The world size used for LMCache keys kv_rank: The kv rank used for LMCache keys vllm_block_size: The block size used in vLLM + tp_size: Tensor-parallel size for MLA + multi-reader locking (default 1). """ self.mq_client = MessageQueueClient(server_url, context) @@ -133,6 +136,7 @@ class LMCacheMPSchedulerAdapter: self.model_name = model_name self.world_size = world_size self.worker_id = kv_rank + self.tp_size = tp_size # Read chunk size from lmcache self.chunk_size = get_lmcache_chunk_size(self.mq_client) @@ -281,6 +285,7 @@ class LMCacheMPSchedulerAdapter: start=start, end=end, request_id=request_id, + tp_size=self.tp_size, ) def _create_hash_key( @@ -293,6 +298,7 @@ class LMCacheMPSchedulerAdapter: worker_id=None, chunk_hash=chunk_hash, request_id=request_id, + tp_size=self.tp_size, ) diff --git a/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_mp_connector.py b/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_mp_connector.py index 38dd980c62..2afdac38c2 100644 --- a/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_mp_connector.py +++ b/vllm/distributed/kv_transfer/kv_connector/v1/lmcache_mp_connector.py @@ -1,6 +1,7 @@ # SPDX-License-Identifier: Apache-2.0 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project import enum +import inspect from collections.abc import Iterable from dataclasses import dataclass, field from typing import TYPE_CHECKING, Any, Literal @@ -52,6 +53,12 @@ if TYPE_CHECKING: logger = lmcache_init_logger(__name__) +def _adapter_accepts_tp_size() -> bool: + """Check if the imported adapter accepts tp_size.""" + sig = inspect.signature(LMCacheMPSchedulerAdapter.__init__) + return "tp_size" in sig.parameters + + # Helper functions def reformat_block_ids(block_ids: tuple[list[int], ...] | None) -> list[int]: if block_ids is None: @@ -101,6 +108,14 @@ def create_scheduler_adapter( vllm_config.parallel_config.rank, vllm_config, ) + tp_size = vllm_config.parallel_config.tensor_parallel_size + + # Pass tp_size only when the adapter accepts it so that + # a newer vllm can still work with an older LMCache. + kwargs: dict[str, Any] = {} + if _adapter_accepts_tp_size(): + kwargs["tp_size"] = tp_size + return LMCacheMPSchedulerAdapter( server_url, zmq_context, @@ -108,6 +123,7 @@ def create_scheduler_adapter( world_size, kv_rank, vllm_config.cache_config.block_size, + **kwargs, )