import os import sys import time import json import queue import argparse import threading from typing import Dict, Any, List from vs_types import GroupConfig, BookerStandaloneConfig, RoleStatus, utc_now_str from gco_wrapper import GCOWrapper from logger_setup import setup_app_logger from configure import REDIS_CFG from sentinel import SentinelGCO from booker import BuiltinBookerGCO, OrderBookerGCO from standalone import StandaloneGCO from utils.safe_redis_cli import SafeRedisClient from utils.cloud_config import ( load_remote_or_cache, compute_config_hash, fetch_cloud_config, save_local_cache, ) from utils.plugin_updater import ( get_installed_plugins, get_installed_plugin_metas, is_newer_version, download_and_install_plugin, ) # 升级指令队列(由 Redis PubSub 线程推送) upgrade_queue = queue.Queue() def get_gco_class(role: str, cfg: Any): """根据角色与配置决定具体的 GCO 实现类""" if role == "sentinel": return SentinelGCO if role == "standalone": return StandaloneGCO if role == "booker": if cfg.booker.account_source == "order": return OrderBookerGCO return BuiltinBookerGCO raise ValueError(f"Unknown role: {role}") def parse_group_configs(role: str, raw_conf: Any) -> Dict[str, Any]: """统一解析组配置为字典 {identifier: config_obj}""" items = raw_conf if isinstance(raw_conf, list) else ([raw_conf] if isinstance(raw_conf, dict) else []) cfg_cls = BookerStandaloneConfig if role == "standalone" else GroupConfig result = {} for item in items: cfg = cfg_cls.from_json(item) result[cfg.identifier] = cfg return result def collect_role_status(role: str, node_id: str, wrappers: Dict[str, GCOWrapper]) -> RoleStatus: """收集当前节点下所有 GroupWrapper 的聚合状态及本地已安装插件完整元数据列表""" groups_status = [wrapper.get_status() for wrapper in wrappers.values()] installed = get_installed_plugin_metas() return RoleStatus( role=role, node_id=node_id, installed_plugins=installed, groups=groups_status, collected_at=utc_now_str(), ) def reload_wrappers_for_plugin( plugin_name: str, wrappers: Dict[str, GCOWrapper], gco_classes: Dict[str, type], logger, ): """找到正在使用该插件的 Wrapper 并进行动态热重载""" for gid, wrapper in list(wrappers.items()): if wrapper.m_cfg.plugin_config.plugin_name == plugin_name: logger.info(f"Hot-reloading group [{gid}] for upgraded plugin '{plugin_name}'...") try: wrapper.stop() except Exception as e: logger.error(f"Error stopping group [{gid}] during reload: {e}") # 清除旧模块缓存以加载最新代码 sys.modules.pop(plugin_name, None) try: target_cls = gco_classes[gid] new_wrapper = GCOWrapper(target_cls, wrapper.m_cfg, REDIS_CFG) new_wrapper.load() new_wrapper.start() wrappers[gid] = new_wrapper logger.info(f"Group [{gid}] successfully reloaded with updated plugin '{plugin_name}'.") except Exception as e: logger.error(f"Failed to restart group [{gid}] after upgrade: {e}") def handle_plugin_upgrades( upgrade_items: List[Dict[str, Any]], wrappers: Dict[str, GCOWrapper], gco_classes: Dict[str, type], logger, ) -> bool: """ 检查并执行插件升级列表 支持格式: [ { "plugin_name": "tls_plugin", "version": "1.1.0", "download_url": "https://...", # 或 "code": "..." "filename": "tls_plugin.py" } ] """ installed = get_installed_plugins() upgraded_any = False for item in upgrade_items: plugin_name = item.get("plugin_name") target_version = str(item.get("version", "")) if not plugin_name or not target_version: continue current_version = installed.get(plugin_name, "0.0.0") if is_newer_version(target_version, current_version): logger.info( f"🚀 Upgrading plugin '{plugin_name}' (v{current_version} -> v{target_version})..." ) try: download_and_install_plugin( plugin_name=plugin_name, version=target_version, download_url=item.get("download_url"), code_content=item.get("code"), filename=item.get("filename"), ) logger.info(f"✅ Plugin '{plugin_name}' v{target_version} installed successfully.") reload_wrappers_for_plugin(plugin_name, wrappers, gco_classes, logger) upgraded_any = True installed[plugin_name] = target_version except Exception as e: logger.error(f"❌ Plugin '{plugin_name}' upgrade failed: {e}") return upgraded_any def start_upgrade_listener(redis_conf: Dict, node_id: str, logger): """启动 Redis PubSub 后台线程,秒级监听云端下发的即时升级指令""" def _listen(): import redis while True: try: r = redis.Redis(**redis_conf) pubsub = r.pubsub() channels = ["vs:channel:upgrade", f"vs:channel:upgrade:{node_id}"] pubsub.subscribe(*channels) logger.info(f"Subscribed to upgrade channels: {channels}") for msg in pubsub.listen(): if msg and msg.get("type") == "message": try: raw_data = msg.get("data") if isinstance(raw_data, bytes): raw_data = raw_data.decode("utf-8") payload = json.loads(raw_data) logger.info(f"Received cloud upgrade command: {payload}") upgrade_queue.put(payload) except Exception as e: logger.error(f"Error parsing upgrade message: {e}") except Exception as e: logger.warning(f"PubSub connection error: {e}. Reconnecting in 5s...") time.sleep(5) threading.Thread(target=_listen, daemon=True, name="Upgrade-Listener").start() def sync_groups( role: str, raw_conf: Any, wrappers: Dict[str, GCOWrapper], gco_classes: Dict[str, type], logger, ): """根据最新配置同步 GCOWrapper 实例的生命周期(停止、重启、更新、新增)""" new_groups = parse_group_configs(role, raw_conf) # 1. 停止已被移除或禁用的组 for gid in list(wrappers.keys()): cfg = new_groups.get(gid) if not cfg or not cfg.enable: logger.info(f"Stopping disabled/removed group [{gid}]...") try: wrappers[gid].stop() except Exception as e: logger.error(f"Error stopping group [{gid}]: {e}") del wrappers[gid] gco_classes.pop(gid, None) # 2. 处理现有组更新或新增组 for gid, cfg in new_groups.items(): if not cfg.enable: continue target_cls = get_gco_class(role, cfg) if gid in wrappers: if gco_classes.get(gid) != target_cls: logger.info(f"Mode changed for group [{gid}]. Restarting with {target_cls.__name__}...") try: wrappers[gid].stop() except Exception as e: logger.error(f"Error stopping group [{gid}]: {e}") try: wrapper = GCOWrapper(target_cls, cfg, REDIS_CFG) wrapper.load() wrapper.start() wrappers[gid] = wrapper gco_classes[gid] = target_cls except Exception as e: logger.error(f"Failed to restart group [{gid}]: {e}") else: logger.info(f"Updating configuration for group [{gid}]...") try: wrappers[gid].update_config(cfg) except Exception as e: logger.error(f"Error updating config for group [{gid}]: {e}") else: logger.info(f"Starting group [{gid}] ({target_cls.__name__})...") try: wrapper = GCOWrapper(target_cls, cfg, REDIS_CFG) wrapper.load() wrapper.start() wrappers[gid] = wrapper gco_classes[gid] = target_cls except Exception as e: logger.error(f"Failed to start group [{gid}]: {e}") def main(): parser = argparse.ArgumentParser(description="Coordinate Unified Runner") parser.add_argument("role", choices=["sentinel", "booker", "standalone"], help="Role to run") parser.add_argument("-c", "--config", type=str, default=None, help="Path to local config file") parser.add_argument("--poll-interval", type=float, default=30.0, help="Cloud config poll interval in seconds") parser.add_argument("--status-interval", type=float, default=60.0, help="RoleStatus reporting interval in seconds (default: 60s)") args = parser.parse_args() role = args.role use_cloud = args.config is None config_path = args.config or f"config/config_{role}.json" node_id = os.getenv("NODE_ID", "node01") config_key = f"{role}_config:{node_id}" status_key = f"vs:status:{role}:{node_id}" logger = setup_app_logger(role.title()) logger.info(f"{role.title()} Runner started (Node ID: '{node_id}', Cloud Mode: {use_cloud})") redis_client = SafeRedisClient(REDIS_CFG, logger.info) # 启动云端指令实时监听(响应云端下发的升级指令) start_upgrade_listener(REDIS_CFG, node_id, logger) groups_conf, current_hash, _ = load_remote_or_cache( config_key=config_key, local_cache_path=config_path, use_cloud=use_cloud, logger=logger.info, ) wrappers: Dict[str, GCOWrapper] = {} gco_classes: Dict[str, type] = {} sync_groups(role, groups_conf, wrappers, gco_classes, logger) logger.info(f"Successfully initialized {len(wrappers)} active group(s).") # 打印本地已安装插件列表与初始 RoleStatus initial_status = collect_role_status(role, node_id, wrappers) status_dict = initial_status.to_dict() logger.info("📊 Initial RoleStatus:\n" + json.dumps(status_dict, indent=2, ensure_ascii=False)) report_ttl = max(180, int(args.status_interval * 3)) try: redis_client.setex(status_key, report_ttl, json.dumps(initial_status.to_dict())) except Exception as e: logger.warning(f"Error reporting initial RoleStatus to Redis: {e}") if use_cloud: logger.info(f"Cloud config active for key '{config_key}'. Polling every {args.poll_interval}s.") last_poll_time = time.time() last_status_report_time = time.time() try: while True: time.sleep(1) now = time.time() # 1. 响应云端下发的即时升级指令 (PubSub) while not upgrade_queue.empty(): try: payload = upgrade_queue.get_nowait() plugins_items = payload.get("plugins", []) if isinstance(plugins_items, dict): plugins_items = [plugins_items] if plugins_items: if handle_plugin_upgrades(plugins_items, wrappers, gco_classes, logger): # 升级成功,立即刷新并上报最新 RoleStatus new_status = collect_role_status(role, node_id, wrappers) status_dict = new_status.to_dict() redis_client.setex(status_key, report_ttl, json.dumps(status_dict)) logger.info("📊 Reported RoleStatus:\n" + json.dumps(status_dict, indent=2, ensure_ascii=False)) except queue.Empty: break # 2. 定时上报包含 installed_plugins 的 RoleStatus if now - last_status_report_time >= args.status_interval: last_status_report_time = now try: current_status = collect_role_status(role, node_id, wrappers) status_dict = current_status.to_dict() redis_client.setex(status_key, report_ttl, json.dumps(status_dict)) logger.info("📊 Reported RoleStatus:\n" + json.dumps(status_dict, indent=2, ensure_ascii=False)) except Exception as e: logger.warning(f"Error reporting RoleStatus to Redis '{status_key}': {e}") # 3. 云端分组配置定时轮询与热更新 if use_cloud and (now - last_poll_time >= args.poll_interval): last_poll_time = now try: new_conf = fetch_cloud_config(config_key) if new_conf is not None: new_hash = compute_config_hash(new_conf) if new_hash != current_hash: logger.info(f"Cloud config change detected ({current_hash[:8]} -> {new_hash[:8]}). Auto-reloading...") save_local_cache(config_path, new_conf) sync_groups(role, new_conf, wrappers, gco_classes, logger) current_hash = new_hash logger.info(f"Auto-reload completed. Currently {len(wrappers)} group(s) running.") except Exception as e: logger.warning(f"Error polling cloud config key '{config_key}': {e}") except KeyboardInterrupt: logger.info(f"Shutting down {role}...") for gid, wrapper in list(wrappers.items()): try: wrapper.stop() except Exception as e: logger.error(f"Error stopping group [{gid}]: {e}") if __name__ == "__main__": main()