文档首页/
AI开发平台ModelArts/
最佳实践/
LLM大语言模型推理/
LLM大语言模型推理历史版本文档/
主流开源大模型基于Lite Server&Cluster适配Ascend-vLLM PyTorch NPU推理指导(6.5.905)/
附录/
ray_worker_checker.py
更新时间:2025-07-29 GMT+08:00
ray_worker_checker.py
ray_worker_checker.py脚本用于等待Ray worker节点连接完成。
在run_vllm_multi_node.sh中会用到,无需修改。
import ray import sys import time def wait_ray_workers_ready(head_addr: str, target_count: int): ray.init(address=head_addr) while True: try: nodes = ray.nodes() if len(nodes) == target_count: print("ray workers are ready.") break else: print("ray workers not ready, check 10s later....") time.sleep(10) except Exception as e: print(e) if __name__ == "__main__": address = sys.argv[1] num_nodes = int(sys.argv[2]) wait_ray_workers_ready(address, num_nodes)
父主题: 附录