[Bugfix] Add custom Triton cache manager to resolve MoE MP issue

vllm/executor/multiproc_gpu_executor.py

@@ -9,6 +9,7 @@
                                                   ResultHandler, WorkerMonitor)
 from vllm.logger import init_logger
 from vllm.sequence import ExecuteModelRequest, SamplerOutput
+from vllm.triton_utils import maybe_set_triton_cache_manager
 from vllm.utils import (cuda_device_count_stateless,
                         error_on_invalid_device_count_status,
                         get_distributed_init_method, get_open_port,
@@ -42,6 +43,10 @@ def _init_executor(self) -> None:
         if "OMP_NUM_THREADS" not in os.environ:
             os.environ["OMP_NUM_THREADS"] = "1"
 
+        # workaround for https://github.com/vllm-project/vllm/issues/6103
+        if world_size > 1:
+            maybe_set_triton_cache_manager()
+
         assert world_size <= cuda_device_count_stateless(), (
             "please set tensor_parallel_size to less than max local gpu count")
 

vllm/triton_utils/__init__.py

@@ -0,0 +1,6 @@
+from vllm.triton_utils.custom_cache_manager import (
+    maybe_set_triton_cache_manager)
+
+__all__ = [
+    "maybe_set_triton_cache_manager",
+]

vllm/triton_utils/custom_cache_manager.py

@@ -0,0 +1,53 @@
+import os
+
+from triton.runtime.cache import (FileCacheManager, default_cache_dir,
+                                  default_dump_dir, default_override_dir)
+
+from vllm.logger import init_logger
+
+logger = init_logger(__name__)
+
+
+def maybe_set_triton_cache_manager() -> None:
+    """Set environment variable to tell Triton to use a
+    custom cache manager"""
+    cache_manger = os.environ.get("TRITON_CACHE_MANAGER", None)
+    if cache_manger is None:
+        manager = "vllm.triton_utils.custom_cache_manager:CustomCacheManager"
+        logger.info("Setting Triton cache manager to: %s", manager)
+        os.environ["TRITON_CACHE_MANAGER"] = manager
+
+
+class CustomCacheManager(FileCacheManager):
+    """Re-implements Triton's cache manager, ensuring that a
+    unique cache directory is created for each process. This is
+    needed to avoid collisions when running with tp>1 and
+    using multi-processing as the distributed backend.
+
+    Note this issue was fixed by triton-lang/triton/pull/4295,
+    but the fix is not yet included in triton==v3.0.0. However,
+    it should be included in the subsequent version.
+    """
+
+    def __init__(self, key, override=False, dump=False):
+        self.key = key
+        self.lock_path = None
+        if dump:
+            self.cache_dir = default_dump_dir()
+            self.cache_dir = os.path.join(self.cache_dir, self.key)
+            self.lock_path = os.path.join(self.cache_dir, "lock")
+            os.makedirs(self.cache_dir, exist_ok=True)
+        elif override:
+            self.cache_dir = default_override_dir()
+            self.cache_dir = os.path.join(self.cache_dir, self.key)
+        else:
+            # create cache directory if it doesn't exist
+            self.cache_dir = os.getenv("TRITON_CACHE_DIR",
+                                       "").strip() or default_cache_dir()
+            if self.cache_dir:
+                self.cache_dir = f"{self.cache_dir}_{os.getpid()}"
+                self.cache_dir = os.path.join(self.cache_dir, self.key)
+                self.lock_path = os.path.join(self.cache_dir, "lock")
+                os.makedirs(self.cache_dir, exist_ok=True)
+            else:
+                raise RuntimeError("Could not create or locate cache dir")