Refactor: Enhance Server Stability and Graceful Shutdown

[BigUncle]

Description (English)

This pull request significantly enhances the server's stability, reliability, and observability by refactoring the startup process and implementing robust error handling. The changes are based on a comprehensive code review process, incorporating multiple rounds of feedback to ensure a production-ready implementation.

Key Changes:

• Robust Initialization: The server initialization logic has been refactored to use an initializationPromise. This ensures that all critical services (like TransformerService, ProviderService) are fully loaded and ready before the server starts listening for incoming requests, completely eliminating potential race conditions at startup.
• Graceful Shutdown on Critical Errors: Implemented graceful shutdown mechanisms for unhandledRejection and uncaughtException events. Instead of an abrupt process exit, the server now attempts to close existing connections gracefully (this.app.close()) before terminating. This minimizes disruption to in-flight requests.
• Enhanced Error Handling Safety:
  • Added .catch() blocks to the graceful shutdown logic to prevent recursive errors if the shutdown process itself fails.
  • Ensured that the reason for an unhandledRejection is always an Error object by wrapping non-Error reasons, which improves type safety and logging consistency.
• Improved Startup Error Logging: Moved the await for the initialization promise into the main try...catch block of the start() method. This allows any initialization failures to be caught and logged with a specific "Error starting server" message, greatly improving diagnostics for startup issues.
• Structured Logging: Refactored log.error calls to use Pino's structured logging format ({ err: error }). This enables the logger to correctly serialize error objects, including stack traces, providing much richer and more useful information for debugging in production environments.
• Code Cleanup: Replaced all debugging console.log statements with this.app.log.debug to align with the project's logging standards and allow for log level control.

These changes collectively make the server more resilient, easier to debug, and safer to operate in a production environment.

---

描述 (简体中文)

本次 Pull Request 通过重构启动流程和实现健壮的错误处理机制, 显著增强了服务器的稳定性、可靠性和可观测性. 所有变更都基于全面的代码审查过程, 融合了多轮反馈, 以确保代码达到生产环境部署标准.

主要变更:

• 健壮的初始化流程: 重构了服务器初始化逻辑, 使用 initializationPromise 来确保所有关键服务 (如 TransformerService, ProviderService) 在服务器开始监听请求之前完全加载就绪, 从根本上消除了启动时的竞态条件风险.
• 关键错误下的优雅关停: 为 unhandledRejection 和 uncaughtException 事件实现了优雅关停机制. 当发生这些严重错误时, 服务器不再突然退出, 而是会先尝试优雅地关闭现有连接 (this.app.close()), 然后再终止进程, 最大限度地减少对正在处理的请求的干扰.
• 增强的错误处理安全性:
  • 在优雅关停逻辑中增加了 .catch() 块, 以防止关停过程本身也抛出异常而导致的无限递归错误.
  • 通过包装非 Error 类型的拒绝原因, 确保了 unhandledRejection 的 reason 始终是一个 Error 对象, 提升了类型安全性和日志记录的一致性.
• 优化的启动错误日志: 将初始化 Promise 的 await 操作移入了 start() 方法的主 try...catch 块内. 这使得任何初始化失败都能被捕获, 并记录为明确的 "Error starting server" 消息, 极大地改善了启动问题的诊断能力.
• 结构化日志: 重构了 log.error 的调用方式, 以使用 Pino 的结构化日志格式 ({ err: error }). 这让日志库能够正确地序列化错误对象 (包括堆栈跟踪), 为生产环境的调试提供了更丰富、更有用的信息.
• 代码清理: 将所有用于调试的 console.log 语句替换为 this.app.log.debug, 以符合项目的日志记录标准, 并便于通过日志级别进行控制.

这些变更共同使服务器更具韧性, 更易于调试, 并且在生产环境中运行更安全.