Update Dockerfile

Dockerfile

@@ -109,7 +109,7 @@ if __name__ == "__main__":
         exit(1)
 EOF
 
-# 复制上传脚本，增强 Gitea 错误日志并在重试上限时抛出错误
+# 复制上传脚本，恢复第一版的错误抛出逻辑，保留北京时间和 Gitea 重试
 COPY --chown=git:git <<'EOF' /uploadhf.py
 import os
 import time
@@ -151,6 +151,7 @@ class DataDirectoryHandler(FileSystemEventHandler):
         self.last_commit_time = 0
         self.commit_delay = 1  # 防抖延迟 1 秒
         self.pending_changes = []  # 缓冲待上传变更
+        self.loop = asyncio.get_event_loop()  # 获取主事件循环
         logger.info(f"初始化监控器，监控目录: {data_directory}，目标仓库: {repo_id}")
     
     def on_any_event(self, event):
@@ -165,8 +166,8 @@ class DataDirectoryHandler(FileSystemEventHandler):
         current_time = time.time()
         if current_time - self.last_commit_time > self.commit_delay:
             self.last_commit_time = current_time
-            # 使用 asyncio.create_task 异步调度提交
-            asyncio.create_task(self.commit_changes(change_type))
+            # 在主事件循环中调度异步任务
+            self.loop.call_soon_threadsafe(lambda: asyncio.create_task(self.commit_changes(change_type)))
 
     async def commit_changes(self, change_type):
         """异步提交变更到 Hugging Face Hub，带重试机制"""
@@ -183,7 +184,7 @@ class DataDirectoryHandler(FileSystemEventHandler):
                     repo_id=self.repo_id,
                     repo_type="dataset",
                     commit_message=commit_message,
-                    ignore_patterns=["*.tmp", "*._workers", "*.log", "*.temp", ".git/*"]
+                    ignore_patterns=["*.tmp", "*.log", "*.temp", ".git/*"]
                 )
                 logger.info(f"✅ 成功提交变更到 Hugging Face Hub: {commit_message}")
                 return
@@ -203,7 +204,8 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
         logger.info(f"HF_TOKEN value: {hf_token}")
     
     if not repo_id:
-        raise ValueError("必须提供 repo_id 参数")
+        logger.error("❌ 必须提供 repo_id 参数")
+        exit(1)
     
     if not os.path.exists(data_directory):
         logger.warning(f"监控目录 {data_directory} 不存在，正在创建...")
@@ -223,92 +225,49 @@ def start_directory_monitoring(data_directory="/data", repo_id=None, hf_token=No
     return observer
 
 def start_gitea_server(port=7860):
-    """启动 Gitea 服务器，包含自动重试逻辑和超时机制，达到重试上限时抛出错误"""
+    """启动 Gitea 服务器"""
     def run_gitea():
         max_restart_attempts = 5
         restart_delay = 10  # 每次重启前的等待时间（秒）
-        startup_timeout = 30  # 启动超时时间（秒）
         attempt = 0
 
-        # 检查 /data 目录权限和数据库状态
-        logger.info("检查 /data 目录权限和 Gitea 数据库状态")
-        os.system("ls -la /data 2>/dev/null")
-        os.system("ls -la /data/gitea 2>/dev/null || echo '/data/gitea 不存在'")
-        if os.path.exists("/data/gitea/gitea.db"):
-            logger.info("Gitea 数据库文件存在: /data/gitea/gitea.db")
-        else:
-            logger.warning("Gitea 数据库文件不存在: /data/gitea/gitea.db")
-
         while attempt < max_restart_attempts:
             try:
-                # 启动 Gitea 进程
                 gitea_process = subprocess.Popen(
                     ['gitea', 'web', '--port', str(port)],
                     stdout=subprocess.PIPE,
                     stderr=subprocess.PIPE,
                     universal_newlines=True
                 )
-                logger.info(f"🚀 Gitea 服务器启动尝试 {attempt + 1}/{max_restart_attempts}，端口: {port}")
+                logger.info(f"🚀 Gitea 服务器已启动，端口: {port}")
                 logger.info(f"📊 访问地址: http://localhost:{port}")
-
-                # 等待 Gitea 启动（超时机制）
-                start_time = time.time()
-                while time.time() - start_time < startup_timeout:
-                    if gitea_process.poll() is not None:
+                
+                while True:
+                    output = gitea_process.stdout.readline()
+                    if output == '' and gitea_process.poll() is not None:
                         break
-                    line = gitea_process.stderr.readline()
-                    if line:
-                        logger.info(f"Gitea stderr: {line.strip()}")
-                    time.sleep(0.1)
-
-                # 检查进程状态
+                    if output:
+                        logger.info(f"Gitea: {output.strip()}")
+                
                 return_code = gitea_process.poll()
-                if return_code is not None:
-                    # 进程已退出
+                if return_code != 0:
                     error_output = gitea_process.stderr.read()
                     logger.error(f"❌ Gitea 服务器异常退出，返回码: {return_code}")
-                    if error_output:
-                        logger.error(f"错误信息: {error_output.strip()}")
+                    logger.error(f"错误信息: {error_output}")
                     attempt += 1
                     if attempt < max_restart_attempts:
                         logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
                         time.sleep(restart_delay)
                     else:
-                        error_msg = f"❌ 达到最大重试次数 ({max_restart_attempts})，Gitea 启动失败"
-                        logger.error(error_msg)
-                        raise RuntimeError(error_msg)
+                        logger.error(f"❌ 达到最大重启次数，Gitea 启动失败")
+                        exit(1)
                 else:
-                    # 进程仍在运行，开始读取输出
-                    logger.info("✅ Gitea 服务器启动成功，开始监控输出")
-                    while True:
-                        output = gitea_process.stdout.readline()
-                        if output == '' and gitea_process.poll() is not None:
-                            break
-                        if output:
-                            logger.info(f"Gitea: {output.strip()}")
-                    
-                    return_code = gitea_process.poll()
-                    if return_code != 0:
-                        error_output = gitea_process.stderr.read()
-                        logger.error(f"❌ Gitea 服务器后期异常退出，返回码: {return_code}")
-                        if error_output:
-                            logger.error(f"错误信息: {error_output.strip()}")
-                        attempt += 1
-                        if attempt < max_restart_attempts:
-                            logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
-                            time.sleep(restart_delay)
-                        else:
-                            error_msg = f"❌ 达到最大重试次数 ({max_restart_attempts})，Gitea 启动失败"
-                            logger.error(error_msg)
-                            raise RuntimeError(error_msg)
-                    else:
-                        logger.info("✅ Gitea 服务器正常退出")
-                        break
+                    logger.info("✅ Gitea 服务器正常退出")
+                    break
                 
             except FileNotFoundError:
-                error_msg = "❌ 未找到 gitea 命令，请确保 Gitea 已正确安装"
-                logger.error(error_msg)
-                raise RuntimeError(error_msg)
+                logger.error("❌ 未找到 gitea 命令，请确保 Gitea 已正确安装")
+                exit(1)
             except Exception as e:
                 logger.error(f"❌ 启动 Gitea 服务器时发生错误: {e}")
                 attempt += 1
@@ -316,9 +275,8 @@ def start_gitea_server(port=7860):
                     logger.info(f"将在 {restart_delay} 秒后尝试重启 Gitea (尝试 {attempt + 1}/{max_restart_attempts})")
                     time.sleep(restart_delay)
                 else:
-                    error_msg = f"❌ 达到最大重试次数 ({max_restart_attempts})，Gitea 启动失败: {e}"
-                    logger.error(error_msg)
-                    raise RuntimeError(error_msg)
+                    logger.error(f"❌ 达到最大重试次数，Gitea 启动失败")
+                    exit(1)
     
     gitea_thread = threading.Thread(target=run_gitea)
     gitea_thread.daemon = True
@@ -358,7 +316,7 @@ async def main():
         
         logger.info("✅ 所有服务已启动完成！")
         logger.info("📁 目录监控: /data → Hugging Face Hub")
-        logger.info(f"🌐 Gitea 服务: http://localhost:{CONFIG['gitea_port']}")
+        logger.info("🌐 Gitea 服务: http://localhost:7860")
         logger.info("🛑 按 Ctrl+C 停止所有服务")
         
         try:
@@ -369,7 +327,7 @@ async def main():
             
     except Exception as e:
         logger.error(f"❌ 启动服务时发生错误: {e}")
-        raise  # 重新抛出异常以确保容器退出
+        exit(1)
     finally:
         if 'observer' in locals():
             observer.stop()