Watchdog库长期运行失效问题求助

我有一个使用watchdog库监控服务器文件夹的Python脚本，当有文件创建时会执行相应处理。该脚本通过nssm注册为系统服务，初始几小时运行正常，但后续虽服务仍处于运行状态却无任何响应，EventViewer及应用日志中均未发现报错或警告信息。

我的代码是基于watchdog库的简单实现，具体如下：

import time
import logging
import signal

from watchdog.observers import Observer
from watchdog.events import FileSystemEventHandler
from logging.handlers import TimedRotatingFileHandler

from project import Program
from config import Config


class MonitorFolder(FileSystemEventHandler):

    def __init__(self, config: Config) -> None:

        self.logfile_path = config.config.get('settings','logfile_path')   
        self.monitor_path = str(config.config.get('settings','monitor_path')).replace(r'\\',r'\\')
        #----------------Logging---------------------------------       
        ##FileHandler
        logger=logging.getLogger()
        logger.setLevel(logging.INFO)    
        handler=TimedRotatingFileHandler(filename = self.logfile_path, when ='D', interval = 1)
        formatter=logging.Formatter('%(levelname)s %(asctime)s [%(name)s.%(funcName)s:%(lineno)d] %(message)s')
        handler.setFormatter(formatter)
        logger.addHandler(handler)
        # ##ConsoleHandler
        console=logging.StreamHandler()
        console.setFormatter (formatter)
        console.setLevel(logging.INFO)
        logging.getLogger().addHandler(console)
        #----------------Logging---------------------------------
    
    def on_created(self, event):
        
        if ('Archive' in event.src_path or 'Error' in event.src_path ):
            return
         
        logging.info(f'Folder monitor received "{event.src_path}" event for "{event.event_type}"')

        time.sleep(5)
        
        prog = Program(event.src_path, config)        
        prog.main()

    

if __name__ == "__main__":

    config = Config()
    event_handler = MonitorFolder(config)
    observer = Observer()
    observer.schedule(event_handler, path = event_handler.monitor_path, recursive = True)

    logging.info("Monitoring started...")

    observer.start()
    try:
        while(True):
            time.sleep(1)
            
    except Exception as e:
        logging.error(e)
        observer.stop()
    observer.join()

此外，应用日志文件中也未记录任何错误。请问有没有人遇到过类似问题？

• 事件处理线程被阻塞：Watchdog的Observer依赖事件线程处理文件变化，你在on_created里直接调用prog.main()，如果这个方法执行耗时久、陷入死循环或者死锁，会直接阻塞事件线程，后续的文件创建事件根本无法被处理。建议把业务逻辑放到单独线程执行：

  import threading
  
  def on_created(self, event):
      if ('Archive' in event.src_path or 'Error' in event.src_path ):
          return
       
      logging.info(f'Folder monitor received "{event.src_path}" event for "{event.event_type}"')
  
      time.sleep(5)
      
      # 用独立线程执行处理逻辑，避免阻塞Observer的事件循环
      thread = threading.Thread(target=lambda: Program(event.src_path, config).main())
      thread.daemon = True
      thread.start()
  

• 日志配置重复添加Handler：你在MonitorFolder的__init__里给根logger重复添加Handler，每次实例化都会加一遍，日志输出时可能触发锁竞争，导致进程卡死。建议把日志配置移到主入口，只初始化一次：

  if __name__ == "__main__":
      config = Config()
      # 提前初始化日志，避免重复添加Handler
      logfile_path = config.config.get('settings','logfile_path')
      logger = logging.getLogger()
      logger.setLevel(logging.INFO)    
      handler = TimedRotatingFileHandler(filename=logfile_path, when='D', interval=1)
      formatter = logging.Formatter('%(levelname)s %(asctime)s [%(name)s.%(funcName)s:%(lineno)d] %(message)s')
      handler.setFormatter(formatter)
      logger.addHandler(handler)
      console = logging.StreamHandler()
      console.setFormatter(formatter)
      console.setLevel(logging.INFO)
      logger.addHandler(console)
  
      event_handler = MonitorFolder(config)
      observer = Observer()
      observer.schedule(event_handler, path=event_handler.monitor_path, recursive=True)
  
      logging.info("Monitoring started...")
      # ... 后续代码不变
  

• 主线程循环方式不合理：你用while True: time.sleep(1)维持主线程，在Windows服务环境下，这种空闲循环可能被系统判定为低优先级进程，甚至触发资源回收导致Observer线程挂起。可以直接用Observer的join()方法替代：

  if __name__ == "__main__":
      # ... 日志和初始化代码
      observer.start()
      try:
          observer.join()  # 直接等待Observer线程结束，无需自定义sleep循环
      except KeyboardInterrupt:
          observer.stop()
      observer.join()
  

• 业务逻辑中的隐性问题：prog.main()里可能存在未捕获的异常、资源泄漏（比如文件句柄、数据库连接未关闭），长时间运行后积累导致进程卡死。建议在prog.main()中添加完整的异常捕获，把错误写入日志；同时检查所有资源是否正确释放。

内容的提问来源于stack exchange，提问作者Sbeastan