A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 05:30 UTC
Disable Ray usage stats collection (#2206)
Woosuk Kwon committed
Dec 20, 2023 at 21:52 UTC
3a4fd5ca59ffaf880137914247bd3cac3c730292
1 file changed
+5
vllm/engine/llm_engine.py
+5
@@ -1,4 +1,5 @@
1
import copy
2
+import os
3
import time
4
from functools import partial
5
from typing import TYPE_CHECKING, Any, Iterable, List, Optional, Tuple, Union
@@ -105,6 +106,10 @@ class LLMEngine:
106
107
# Create the parallel GPU workers.
108
if self.parallel_config.worker_use_ray:
109
+ # Disable Ray usage stats collection.
110
+ ray_usage = os.environ.get("RAY_USAGE_STATS_ENABLED", "0")
111
+ if ray_usage != "1":
112
+ os.environ["RAY_USAGE_STATS_ENABLED"] = "0"
113
self._init_workers_ray(placement_group)
114
else:
115
self._init_workers(distributed_init_method)