返回 Skill 列表
extension
分类: 开发与工程无需 API Key

spring-retry

Spring Retry为Spring应用程序提供透明的重试支持。涵盖@Retryable、@Recover、RetryTemplate、退避策略和断路器。使用场景:当用户提到“spring retry”、“@Retryable”、“@Recover”、“RetryTemplate”、“backoff policy”、“retry transient failure”、“automatic retry”。不适用于:断路器模式-应使用`spring-cloud-circuitbreaker`技能,消息重试-应使用`spring-kafka`或消息处理技能中的重试主题

person作者: jakexiaohubgithub

Spring Retry - Quick Reference

Full Reference: See advanced.md for RetryTemplate patterns, custom retry policies, retry listeners, stateful retry, async retry, and testing.

Deep Knowledge: Use mcp__documentation__fetch_docs with technology: spring-retry for comprehensive documentation.

Dependencies

<dependency>
    <groupId>org.springframework.retry</groupId>
    <artifactId>spring-retry</artifactId>
</dependency>
<dependency>
    <groupId>org.springframework</groupId>
    <artifactId>spring-aspects</artifactId>
</dependency>

Enable Retry

@SpringBootApplication
@EnableRetry
public class Application {
    public static void main(String[] args) {
        SpringApplication.run(Application.class, args);
    }
}

@Retryable Annotation

Basic Usage

@Retryable(maxAttempts = 3)
public String callExternalApi(String endpoint) {
    return restClient.get(endpoint);
}

With Specific Exceptions

@Retryable(
    retryFor = {ServiceUnavailableException.class, TimeoutException.class},
    noRetryFor = {PaymentDeclinedException.class},
    maxAttempts = 5
)
public PaymentResult processPayment(PaymentRequest request) {
    return paymentGateway.charge(request);
}

With Backoff

@Retryable(
    maxAttempts = 4,
    backoff = @Backoff(
        delay = 1000,       // Initial delay: 1 second
        multiplier = 2,     // Exponential: 1s, 2s, 4s
        maxDelay = 10000    // Max delay: 10 seconds
    )
)
public void sendNotification(Notification notification) {
    notificationClient.send(notification);
}

@Recover
public void recoverNotification(Exception e, Notification notification) {
    log.error("Failed to send notification after retries: {}", notification.getId());
    deadLetterQueue.add(notification);
}

@Recover Method

@Service
public class UserService {

    @Retryable(retryFor = ServiceException.class, maxAttempts = 3)
    public User getUser(Long id) {
        return userClient.findById(id);
    }

    // Must have same return type
    // First parameter can be the exception
    @Recover
    public User recoverGetUser(ServiceException e, Long id) {
        log.warn("Falling back to cached user for id: {}", id);
        return userCache.get(id);
    }

    // Multiple recover methods for different exceptions
    @Recover
    public User recoverGetUserTimeout(TimeoutException e, Long id) {
        return User.unknown(id);
    }
}

Best Practices

| Do | Don't | |----|-------| | Use exponential backoff | Fixed rapid retries | | Set max attempts limit | Retry indefinitely | | Handle non-retryable exceptions | Retry business errors | | Log retry attempts | Silent retries | | Implement recovery fallback | Let retries fail silently |

When NOT to Use This Skill

  • Circuit breaker - Use spring-cloud-circuitbreaker or Resilience4j
  • Message retry - Use Kafka retry topics or DLT
  • Non-idempotent operations - Ensure idempotency first
  • Business errors - Only retry transient failures

Anti-Patterns

| Anti-Pattern | Problem | Solution | |--------------|---------|----------| | Retry indefinitely | Hangs forever | Set max attempts | | Fixed rapid retry | Overwhelms service | Use exponential backoff | | Retrying business errors | Wastes resources | Use noRetryFor | | No recovery method | Silent failures | Implement @Recover |

Quick Troubleshooting

| Problem | Diagnostic | Fix | |---------|------------|-----| | Retry not happening | Check @EnableRetry | Add to config class | | @Recover not called | Check method signature | Match return type and params | | Too many retries | Check maxAttempts | Reduce or add timeout | | Backoff not working | Check annotation | Verify @Backoff config |

Production Checklist

  • [ ] Appropriate max attempts set
  • [ ] Exponential backoff configured
  • [ ] Max delay capped
  • [ ] Non-retryable exceptions defined
  • [ ] Recovery methods implemented
  • [ ] Retry listeners for monitoring
  • [ ] Metrics on retry counts

Reference Documentation