降级与熔断处理

这里以 Java 和 Spring Boot 为例进行说明。

服务降级
在服务层实现降级逻辑,当上游服务出现异常时,快速返回一个预设的降级响应:

@Service
public class OrderService {
   

    @Autowired
    private PaymentService paymentService;

    public OrderResponse placeOrder(OrderRequest request) {
   
        try {
   
            // 调用支付服务进行支付处理
            paymentService.doPayment(request.getOrderInfo());
            // 其他订单处理逻辑
            // ...
            return new OrderResponse(OrderStatus.PLACED, "Order placed successfully");
        } catch (Exception e) {
   
            // 服务降级逻辑
            log.error("Error occurred while placing order: {}", e.getMessage());
            return new OrderResponse(OrderStatus.FAILED, "Failed to place order, please try again later");
        }
    }
}

熔断处理
使用 Spring Cloud Circuit Breaker 框架实现熔断机制:

@Service
public class PaymentService {
   

    @CircuitBreaker(name = "paymentService", fallbackMethod = "paymentFallback")
    public PaymentResponse doPayment(PaymentInfo paymentInfo) {
   
        // 实际的支付处理逻辑
        return new PaymentResponse(PaymentStatus.SUCCESS, "Payment successful");
    }

    public PaymentResponse paymentFallback(PaymentInfo paymentInfo, Throwable t) {
   
        // 熔断后的降级逻辑
        log.error("Payment service failed: {}", t.getMessage());
        return new PaymentResponse(PaymentStatus.FAILED, "Payment failed, please try again later");
    }
}

在上面的例子中,我们使用 @CircuitBreaker 注解标记了 doPayment 方法,并指定了 fallbackMethod 为 paymentFallback。当 doPayment 方法抛出异常时,会自动调用 paymentFallback 方法,返回一个预设的降级响应。

这里还需要在应用程序的配置文件中配置 Circuit Breaker 的相关参数,如熔断阈值、时间窗口、恢复时间等:

resilience4j.circuitbreaker:
  instances:
    paymentService:
      sliding-window-size: 10
      minimum-number-of-calls: 5
      permissible-number-of-calls-in-half-open-state: 3
      automatic-transition-from-open-to-half-open-state-enabled: true
      wait-duration-in-open-state: 10s

通过以上的降级和熔断处理,我们可以在上游服务出现异常时,快速返回一个友好的响应,保护系统不被拖垮,同时也能给用户一个更好的体验。