当前位置：首页 > ops >正文

Spring如何解决循环依赖：深入理解三级缓存机制

ops 2025/9/6 6:25:34

Spring如何解决循环依赖：深入理解三级缓存机制

引言

在我们之前的文章中，我们探讨了什么是循环依赖以及它带来的问题。作为Java生态系统中最重要的框架之一，Spring Framework在处理循环依赖方面有着独特而精妙的解决方案。今天，让我们深入了解Spring是如何通过三级缓存机制巧妙地解决循环依赖问题的。

Spring中的循环依赖场景

典型的循环依赖示例

@Component
public class ServiceA {@Autowiredprivate ServiceB serviceB;public void doSomething() {System.out.println("ServiceA doing something");serviceB.doSomething();}
}@Component
public class ServiceB {@Autowiredprivate ServiceC serviceC;public void doSomething() {System.out.println("ServiceB doing something");serviceC.doSomething();}
}@Component
public class ServiceC {@Autowiredprivate ServiceA serviceA;  // 形成循环依赖：A -> B -> C -> Apublic void doSomething() {System.out.println("ServiceC doing something");serviceA.doSomething();}
}

构造器循环依赖（Spring无法解决）

@Component
public class ServiceA {private final ServiceB serviceB;// 构造器注入形成的循环依赖，Spring无法解决public ServiceA(ServiceB serviceB) {this.serviceB = serviceB;}
}@Component
public class ServiceB {private final ServiceA serviceA;public ServiceB(ServiceA serviceA) {this.serviceA = serviceA;}
}

Spring Bean的生命周期

要理解Spring如何解决循环依赖，首先需要了解Bean的创建过程：

public class SpringBeanLifecycle {/*** Spring Bean的完整生命周期*/public Object createBean(String beanName, BeanDefinition beanDefinition) {// 1. 实例化 - 创建对象实例（调用构造器）Object bean = instantiateBean(beanName, beanDefinition);// 2. 属性填充 - 依赖注入populateBean(bean, beanName, beanDefinition);// 3. 初始化 - 调用初始化方法initializeBean(bean, beanName);return bean;}private Object instantiateBean(String beanName, BeanDefinition bd) {// 通过反射创建实例Constructor<?> constructor = bd.getBeanClass().getDeclaredConstructor();return constructor.newInstance();}private void populateBean(Object bean, String beanName, BeanDefinition bd) {// 处理@Autowired、@Resource等注解// 这里会触发依赖Bean的创建Field[] fields = bean.getClass().getDeclaredFields();for (Field field : fields) {if (field.isAnnotationPresent(Autowired.class)) {Object dependency = getBean(field.getType());field.setAccessible(true);field.set(bean, dependency);}}}private void initializeBean(Object bean, String beanName) {// 调用初始化回调方法if (bean instanceof InitializingBean) {((InitializingBean) bean).afterPropertiesSet();}}
}

Spring的三级缓存机制

Spring通过三级缓存来解决单例Bean的循环依赖问题：

public class DefaultSingletonBeanRegistry {/** 一级缓存：完成初始化的单例Bean */private final Map<String, Object> singletonObjects = new ConcurrentHashMap<>(256);/** 二级缓存：完成实例化但未完成初始化的Bean */private final Map<String, Object> earlySingletonObjects = new HashMap<>(16);/** 三级缓存：单例Bean的工厂 */private final Map<String, ObjectFactory<?>> singletonFactories = new HashMap<>(16);/** 正在创建的Bean名称集合 */private final Set<String> singletonsCurrentlyInCreation = Collections.newSetFromMap(new ConcurrentHashMap<>(16));/*** 获取单例Bean的核心方法*/protected Object getSingleton(String beanName) {return getSingleton(beanName, true);}protected Object getSingleton(String beanName, boolean allowEarlyReference) {// 1. 从一级缓存中获取完全初始化的BeanObject singletonObject = this.singletonObjects.get(beanName);if (singletonObject == null && isSingletonCurrentlyInCreation(beanName)) {synchronized (this.singletonObjects) {// 2. 从二级缓存中获取早期Bean引用singletonObject = this.earlySingletonObjects.get(beanName);if (singletonObject == null && allowEarlyReference) {// 3. 从三级缓存中获取Bean工厂ObjectFactory<?> singletonFactory = this.singletonFactories.get(beanName);if (singletonFactory != null) {// 调用工厂方法创建早期Bean引用singletonObject = singletonFactory.getObject();// 将早期Bean引用放入二级缓存this.earlySingletonObjects.put(beanName, singletonObject);// 从三级缓存中移除this.singletonFactories.remove(beanName);}}}}return singletonObject;}/*** 添加单例Bean工厂到三级缓存*/protected void addSingletonFactory(String beanName, ObjectFactory<?> singletonFactory) {synchronized (this.singletonObjects) {if (!this.singletonObjects.containsKey(beanName)) {this.singletonFactories.put(beanName, singletonFactory);this.earlySingletonObjects.remove(beanName);}}}/*** 将完全初始化的Bean添加到一级缓存*/protected void addSingleton(String beanName, Object singletonObject) {synchronized (this.singletonObjects) {this.singletonObjects.put(beanName, singletonObject);this.singletonFactories.remove(beanName);this.earlySingletonObjects.remove(beanName);}}
}

循环依赖解决过程详解

让我们通过一个具体的例子来看看Spring是如何解决循环依赖的：

/*** 模拟Spring解决循环依赖的完整过程*/
public class CircularDependencyResolver {// 三级缓存private Map<String, Object> singletonObjects = new ConcurrentHashMap<>();private Map<String, Object> earlySingletonObjects = new HashMap<>();private Map<String, ObjectFactory<?>> singletonFactories = new HashMap<>();private Set<String> singletonsCurrentlyInCreation = new HashSet<>();/*** 创建Bean的主要方法*/public Object getBean(String beanName) {// 先尝试从缓存中获取Object singleton = getSingleton(beanName);if (singleton != null) {return singleton;}// 缓存中没有，开始创建return createBean(beanName);}private Object createBean(String beanName) {// 标记Bean正在创建中singletonsCurrentlyInCreation.add(beanName);try {// 1. 实例化BeanObject bean = instantiateBean(beanName);// 2. 将Bean工厂放入三级缓存（关键步骤）addSingletonFactory(beanName, () -> getEarlyBeanReference(beanName, bean));// 3. 属性填充（可能触发循环依赖）populateBean(bean, beanName);// 4. 初始化BeaninitializeBean(bean, beanName);// 5. 将完成的Bean放入一级缓存addSingleton(beanName, bean);return bean;} finally {// 移除创建中标记singletonsCurrentlyInCreation.remove(beanName);}}private Object instantiateBean(String beanName) {// 模拟通过反射创建Bean实例System.out.println("实例化Bean: " + beanName);if ("serviceA".equals(beanName)) {return new ServiceA();} else if ("serviceB".equals(beanName)) {return new ServiceB();}throw new RuntimeException("Unknown bean: " + beanName);}private void populateBean(Object bean, String beanName) {System.out.println("填充Bean属性: " + beanName);// 模拟依赖注入if (bean instanceof ServiceA) {ServiceA serviceA = (ServiceA) bean;// 注入ServiceB，这里会触发ServiceB的创建Object serviceB = getBean("serviceB");serviceA.setServiceB((ServiceB) serviceB);} else if (bean instanceof ServiceB) {ServiceB serviceB = (ServiceB) bean;// 注入ServiceA，这里会从缓存中获取早期引用Object serviceA = getBean("serviceA");serviceB.setServiceA((ServiceA) serviceA);}}private void initializeBean(Object bean, String beanName) {System.out.println("初始化Bean: " + beanName);// 执行初始化逻辑}/*** 获取早期Bean引用（处理AOP代理）*/private Object getEarlyBeanReference(String beanName, Object bean) {System.out.println("获取早期Bean引用: " + beanName);// 如果需要AOP代理，在这里创建代理对象if (needsProxy(beanName)) {return createProxy(bean);}return bean;}private boolean needsProxy(String beanName) {// 简化的代理判断逻辑return beanName.contains("service");}private Object createProxy(Object target) {// 简化的代理创建逻辑System.out.println("创建代理对象: " + target.getClass().getSimpleName());return target; // 实际应该返回代理对象}// 其他辅助方法...private Object getSingleton(String beanName) {Object singletonObject = singletonObjects.get(beanName);if (singletonObject == null && singletonsCurrentlyInCreation.contains(beanName)) {singletonObject = earlySingletonObjects.get(beanName);if (singletonObject == null) {ObjectFactory<?> singletonFactory = singletonFactories.get(beanName);if (singletonFactory != null) {singletonObject = singletonFactory.getObject();earlySingletonObjects.put(beanName, singletonObject);singletonFactories.remove(beanName);}}}return singletonObject;}private void addSingletonFactory(String beanName, ObjectFactory<?> singletonFactory) {if (!singletonObjects.containsKey(beanName)) {singletonFactories.put(beanName, singletonFactory);earlySingletonObjects.remove(beanName);}}private void addSingleton(String beanName, Object singletonObject) {singletonObjects.put(beanName, singletonObject);singletonFactories.remove(beanName);earlySingletonObjects.remove(beanName);}
}

循环依赖解决的时序图

ServiceA创建过程:
1. 实例化ServiceA
2. 将ServiceA工厂放入三级缓存
3. 填充ServiceA属性，需要ServiceB|├─ ServiceB创建过程:│  1. 实例化ServiceB  │  2. 将ServiceB工厂放入三级缓存│  3. 填充ServiceB属性，需要ServiceA│  4. 从三级缓存获取ServiceA早期引用│  5. 完成ServiceB初始化，放入一级缓存│
4. 获得ServiceB实例，完成ServiceA属性填充
5. 完成ServiceA初始化，放入一级缓存

为什么需要三级缓存？

只有一级缓存的问题

public class SingleCacheExample {private Map<String, Object> singletonObjects = new ConcurrentHashMap<>();public Object getBean(String beanName) {Object bean = singletonObjects.get(beanName);if (bean != null) {return bean;}// 创建Beanbean = createBean(beanName);singletonObjects.put(beanName, bean);return bean;}// 问题：在Bean完全创建完成之前，其他Bean无法获取到它的引用// 导致循环依赖无法解决
}

只有二级缓存的问题

public class TwoCacheExample {private Map<String, Object> singletonObjects = new ConcurrentHashMap<>();private Map<String, Object> earlySingletonObjects = new HashMap<>();// 问题：无法处理AOP代理的情况// 如果Bean需要被代理，早期引用和最终对象可能不是同一个实例
}

三级缓存的优势

/*** 三级缓存解决的问题：* 1. 一级缓存：存储完全初始化的Bean* 2. 二级缓存：存储早期Bean引用，解决循环依赖* 3. 三级缓存：存储Bean工厂，延迟决定是否需要代理*/
public class ThreeCacheAdvantages {/*** 三级缓存的关键作用：* 1. 延迟代理对象的创建* 2. 确保循环依赖中代理对象的一致性* 3. 避免不必要的代理对象创建*/private Object getEarlyBeanReference(String beanName, Object bean) {// 只有在真正需要早期引用时才决定是否创建代理if (hasCircularDependency(beanName)) {if (needsProxy(beanName)) {return createProxy(bean);}}return bean;}
}

AOP与循环依赖

Spring AOP为循环依赖的解决增加了复杂性：

@Component
@Service
public class UserService {@Autowiredprivate OrderService orderService;@Transactional  // 这个注解会导致创建AOP代理public void createUser(User user) {// 业务逻辑orderService.createDefaultOrder(user);}
}@Component
@Service  
public class OrderService {@Autowiredprivate UserService userService;@Transactionalpublic void createDefaultOrder(User user) {// 业务逻辑userService.validateUser(user);}
}

AOP代理处理过程：

public class AopCircularDependencyHandler {/*** 处理AOP代理的循环依赖*/public Object resolveAopCircularDependency(String beanName, Object bean) {// 1. 检查是否需要创建代理if (shouldCreateProxy(bean)) {// 2. 创建代理对象Object proxy = createAopProxy(bean);// 3. 确保早期引用和最终对象的一致性ensureProxyConsistency(beanName, bean, proxy);return proxy;}return bean;}private boolean shouldCreateProxy(Object bean) {// 检查是否有@Transactional、@Async等注解Class<?> beanClass = bean.getClass();// 检查类级别注解if (beanClass.isAnnotationPresent(Transactional.class) ||beanClass.isAnnotationPresent(Async.class)) {return true;}// 检查方法级别注解Method[] methods = beanClass.getDeclaredMethods();for (Method method : methods) {if (method.isAnnotationPresent(Transactional.class) ||method.isAnnotationPresent(Async.class)) {return true;}}return false;}private Object createAopProxy(Object target) {// 使用CGLIB或JDK动态代理创建代理对象ProxyFactory proxyFactory = new ProxyFactory(target);// 添加事务拦截器proxyFactory.addAdvice(new TransactionInterceptor());return proxyFactory.getProxy();}
}

Spring无法解决的循环依赖场景

1. 构造器循环依赖

@Component
public class ServiceA {private final ServiceB serviceB;// Spring无法解决构造器循环依赖public ServiceA(ServiceB serviceB) {this.serviceB = serviceB;}
}@Component
public class ServiceB {private final ServiceA serviceA;public ServiceB(ServiceA serviceA) {this.serviceA = serviceA;}
}// 启动时会抛出异常：
// BeanCurrentlyInCreationException: Error creating bean with name 'serviceA': 
// Requested bean is currently in creation: Is there an unresolvable circular reference?

2. prototype作用域的循环依赖

@Component
@Scope("prototype")
public class PrototypeServiceA {@Autowiredprivate PrototypeServiceB serviceB;
}@Component
@Scope("prototype")
public class PrototypeServiceB {@Autowiredprivate PrototypeServiceA serviceA;
}// prototype作用域的Bean每次都会创建新实例，无法使用缓存解决循环依赖

3. @Async注解的特殊情况

@Component
public class AsyncServiceA {@Autowiredprivate AsyncServiceB serviceB;@Asyncpublic void doAsync() {serviceB.doSomething();}
}@Component
public class AsyncServiceB {@Autowiredprivate AsyncServiceA serviceA;public void doSomething() {serviceA.doAsync();}
}// 在某些情况下，@Async可能导致循环依赖解决失败

循环依赖的解决方案

1. 使用@Lazy注解

@Component
public class ServiceA {private final ServiceB serviceB;// 使用@Lazy延迟注入，打破循环依赖public ServiceA(@Lazy ServiceB serviceB) {this.serviceB = serviceB;}
}@Component
public class ServiceB {private final ServiceA serviceA;public ServiceB(ServiceA serviceA) {this.serviceA = serviceA;}
}

2. 使用Setter注入替代构造器注入

@Component
public class ServiceA {private ServiceB serviceB;// 使用Setter注入替代构造器注入@Autowiredpublic void setServiceB(ServiceB serviceB) {this.serviceB = serviceB;}
}@Component
public class ServiceB {private ServiceA serviceA;@Autowiredpublic void setServiceA(ServiceA serviceA) {this.serviceA = serviceA;}
}

3. 使用@PostConstruct

@Component
public class ServiceA {@Autowiredprivate ServiceB serviceB;private ServiceA selfReference;@PostConstructpublic void init() {// 在初始化阶段设置自引用this.selfReference = this;}
}

4. 重新设计架构

// 原始设计：循环依赖
@Component
public class UserService {@Autowiredprivate OrderService orderService;public void processUser(User user) {orderService.processOrder(user.getOrders());}
}@Component
public class OrderService {@Autowiredprivate UserService userService;public void processOrder(List<Order> orders) {for (Order order : orders) {userService.validateUser(order.getUser());}}
}// 重构后：提取公共服务
@Component
public class ValidationService {public void validateUser(User user) {// 用户验证逻辑}public void validateOrder(Order order) {// 订单验证逻辑}
}@Component
public class UserService {@Autowiredprivate ValidationService validationService;public void processUser(User user) {validationService.validateUser(user);// 处理用户逻辑}
}@Component
public class OrderService {@Autowiredprivate ValidationService validationService;public void processOrder(Order order) {validationService.validateOrder(order);// 处理订单逻辑}
}

循环依赖检测和调试

1. 启用循环依赖检测

@Configuration
@EnableAutoConfiguration
public class AppConfig {@Beanpublic static BeanFactoryPostProcessor circularDependencyDetector() {return new BeanFactoryPostProcessor() {@Overridepublic void postProcessBeanFactory(ConfigurableListableBeanFactory beanFactory) {if (beanFactory instanceof DefaultListableBeanFactory) {DefaultListableBeanFactory factory = (DefaultListableBeanFactory) beanFactory;factory.setAllowCircularReferences(false); // 禁用循环依赖}}};}
}

2. 自定义循环依赖检测器

@Component
public class CircularDependencyDetector implements BeanPostProcessor {private final Set<String> beansInCreation = new HashSet<>();private final Map<String, Set<String>> dependencyGraph = new HashMap<>();@Overridepublic Object postProcessBeforeInitialization(Object bean, String beanName) {detectCircularDependency(beanName);return bean;}private void detectCircularDependency(String beanName) {if (beansInCreation.contains(beanName)) {throw new BeanCurrentlyInCreationException("Circular dependency detected: " + beanName);}beansInCreation.add(beanName);// 分析依赖关系analyzeDependencies(beanName);beansInCreation.remove(beanName);}private void analyzeDependencies(String beanName) {// 通过反射分析Bean的依赖关系// 构建依赖图，检测环路}
}

3. 循环依赖调试工具

@Component
public class CircularDependencyDebugger {private static final Logger logger = LoggerFactory.getLogger(CircularDependencyDebugger.class);@EventListenerpublic void handleContextRefreshed(ContextRefreshedEvent event) {ApplicationContext context = event.getApplicationContext();// 分析所有Bean的依赖关系analyzeBeanDependencies(context);}private void analyzeBeanDependencies(ApplicationContext context) {String[] beanNames = context.getBeanDefinitionNames();for (String beanName : beanNames) {try {BeanDefinition beanDefinition = ((ConfigurableApplicationContext) context).getBeanFactory().getBeanDefinition(beanName);// 分析依赖关系Set<String> dependencies = extractDependencies(beanDefinition);logger.info("Bean {} depends on: {}", beanName, dependencies);} catch (Exception e) {logger.warn("Failed to analyze dependencies for bean: {}", beanName, e);}}}private Set<String> extractDependencies(BeanDefinition beanDefinition) {Set<String> dependencies = new HashSet<>();// 提取构造器依赖ConstructorArgumentValues constructorArgs = beanDefinition.getConstructorArgumentValues();for (ConstructorArgumentValues.ValueHolder valueHolder : constructorArgs.getGenericArgumentValues()) {if (valueHolder.getValue() instanceof RuntimeBeanReference) {RuntimeBeanReference ref = (RuntimeBeanReference) valueHolder.getValue();dependencies.add(ref.getBeanName());}}// 提取属性依赖MutablePropertyValues propertyValues = beanDefinition.getPropertyValues();for (PropertyValue pv : propertyValues.getPropertyValues()) {if (pv.getValue() instanceof RuntimeBeanReference) {RuntimeBeanReference ref = (RuntimeBeanReference) pv.getValue();dependencies.add(ref.getBeanName());}}return dependencies;}
}

性能考虑

1. 三级缓存的性能影响

public class CachePerformanceAnalysis {/*** 分析三级缓存对性能的影响*/public void analyzePerformance() {// 1. 内存开销：三个Map需要额外的内存// 2. 查找开销：需要依次查找三个缓存// 3. 同步开销：缓存操作需要同步long startTime = System.nanoTime();// 模拟缓存查找Object bean = getSingletonFromCache("testBean");long endTime = System.nanoTime();long duration = endTime - startTime;System.out.println("缓存查找耗时: " + duration + " 纳秒");}private Object getSingletonFromCache(String beanName) {// 三级缓存查找的性能开销Object singleton = singletonObjects.get(beanName);if (singleton == null) {singleton = earlySingletonObjects.get(beanName);if (singleton == null) {ObjectFactory<?> factory = singletonFactories.get(beanName);if (factory != null) {singleton = factory.getObject();}}}return singleton;}
}

2. 优化建议

@Configuration
public class CircularDependencyOptimization {/*** 优化循环依赖处理的配置*/@Beanpublic static BeanFactoryPostProcessor optimizeCircularDependency() {return beanFactory -> {if (beanFactory instanceof DefaultListableBeanFactory) {DefaultListableBeanFactory factory = (DefaultListableBeanFactory) beanFactory;// 1. 预分析Bean依赖关系，提前发现循环依赖preAnalyzeDependencies(factory);// 2. 优化缓存大小optimizeCacheSize(factory);// 3. 启用并发优化enableConcurrentOptimization(factory);}};}```javaprivate static void preAnalyzeDependencies(DefaultListableBeanFactory factory) {String[] beanNames = factory.getBeanDefinitionNames();Map<String, Set<String>> dependencyGraph = new HashMap<>();// 构建依赖图for (String beanName : beanNames) {BeanDefinition bd = factory.getBeanDefinition(beanName);Set<String> dependencies = extractBeanDependencies(bd);dependencyGraph.put(beanName, dependencies);}// 检测循环依赖detectCycles(dependencyGraph);}private static void detectCycles(Map<String, Set<String>> graph) {Set<String> visited = new HashSet<>();Set<String> recursionStack = new HashSet<>();for (String node : graph.keySet()) {if (hasCycle(graph, node, visited, recursionStack)) {System.out.println("检测到循环依赖，涉及Bean: " + node);}}}private static boolean hasCycle(Map<String, Set<String>> graph, String node, Set<String> visited, Set<String> recursionStack) {if (recursionStack.contains(node)) {return true; // 发现环路}if (visited.contains(node)) {return false; // 已经访问过，无环路}visited.add(node);recursionStack.add(node);Set<String> neighbors = graph.get(node);if (neighbors != null) {for (String neighbor : neighbors) {if (hasCycle(graph, neighbor, visited, recursionStack)) {return true;}}}recursionStack.remove(node);return false;}private static void optimizeCacheSize(DefaultListableBeanFactory factory) {// 根据Bean数量优化缓存初始容量int beanCount = factory.getBeanDefinitionCount();int optimalCacheSize = Math.max(16, beanCount / 4);// 这里可以通过反射设置缓存的初始容量// 实际实现中需要访问私有字段}private static void enableConcurrentOptimization(DefaultListableBeanFactory factory) {// 启用并发创建Bean的优化factory.setAllowEagerClassLoading(true);factory.setAllowRawInjectionDespiteWrapping(true);}
}

实际应用案例

案例1：微服务中的循环依赖

/*** 微服务架构中常见的循环依赖场景*/
@Service
public class UserService {@Autowiredprivate OrderService orderService;@Autowiredprivate NotificationService notificationService;@Transactionalpublic User createUser(CreateUserRequest request) {User user = new User(request);user = userRepository.save(user);// 创建用户后，创建默认订单orderService.createWelcomeOrder(user);// 发送欢迎通知notificationService.sendWelcomeNotification(user);return user;}
}@Service
public class OrderService {@Autowiredprivate UserService userService; // 循环依赖@Autowiredprivate PaymentService paymentService;@Transactionalpublic Order createWelcomeOrder(User user) {Order order = new Order(user, getWelcomeProducts());order = orderRepository.save(order);// 更新用户的订单统计userService.updateOrderStatistics(user.getId());return order;}
}@Service
public class NotificationService {@Autowiredprivate UserService userService; // 另一个循环依赖public void sendWelcomeNotification(User user) {// 获取用户偏好设置UserPreference preference = userService.getUserPreference(user.getId());if (preference.isEmailEnabled()) {emailService.sendWelcomeEmail(user);}if (preference.isSmsEnabled()) {smsService.sendWelcomeSms(user);}}
}

解决方案：

/*** 重构后的解决方案：使用事件驱动架构*/
@Service
public class UserService {@Autowiredprivate ApplicationEventPublisher eventPublisher;@Transactionalpublic User createUser(CreateUserRequest request) {User user = new User(request);user = userRepository.save(user);// 发布用户创建事件，而不是直接调用其他服务eventPublisher.publishEvent(new UserCreatedEvent(user));return user;}public void updateOrderStatistics(Long userId) {// 更新用户订单统计User user = userRepository.findById(userId);user.incrementOrderCount();userRepository.save(user);}public UserPreference getUserPreference(Long userId) {return userPreferenceRepository.findByUserId(userId);}
}@Service
public class OrderService {// 移除对UserService的直接依赖@EventListener@Asyncpublic void handleUserCreated(UserCreatedEvent event) {createWelcomeOrder(event.getUser());}@Transactionalpublic Order createWelcomeOrder(User user) {Order order = new Order(user, getWelcomeProducts());order = orderRepository.save(order);// 发布订单创建事件eventPublisher.publishEvent(new OrderCreatedEvent(order));return order;}
}@Service
public class NotificationService {// 移除对UserService的直接依赖@Autowiredprivate UserPreferenceRepository userPreferenceRepository;@EventListener@Asyncpublic void handleUserCreated(UserCreatedEvent event) {sendWelcomeNotification(event.getUser());}public void sendWelcomeNotification(User user) {// 直接查询用户偏好，避免循环依赖UserPreference preference = userPreferenceRepository.findByUserId(user.getId());if (preference.isEmailEnabled()) {emailService.sendWelcomeEmail(user);}if (preference.isSmsEnabled()) {smsService.sendWelcomeSms(user);}}
}// 事件定义
public class UserCreatedEvent extends ApplicationEvent {private final User user;public UserCreatedEvent(Object source, User user) {super(source);this.user = user;}public User getUser() {return user;}
}

案例2：复杂的业务场景循环依赖

/*** 电商系统中的复杂循环依赖场景*/
@Service
public class ProductService {@Autowiredprivate CategoryService categoryService;@Autowiredprivate InventoryService inventoryService;@Autowiredprivate PriceService priceService;public ProductDTO getProductDetails(Long productId) {Product product = productRepository.findById(productId);// 获取分类信息Category category = categoryService.getCategoryWithProducts(product.getCategoryId());// 获取库存信息Inventory inventory = inventoryService.getInventoryWithProductInfo(productId);// 获取价格信息Price price = priceService.getPriceWithProductDetails(productId);return ProductDTO.builder().product(product).category(category).inventory(inventory).price(price).build();}
}@Service
public class CategoryService {@Autowiredprivate ProductService productService; // 循环依赖public Category getCategoryWithProducts(Long categoryId) {Category category = categoryRepository.findById(categoryId);// 获取分类下的热门产品List<ProductDTO> hotProducts = productService.getHotProductsByCategory(categoryId);category.setHotProducts(hotProducts);return category;}
}@Service
public class InventoryService {@Autowiredprivate ProductService productService; // 循环依赖public Inventory getInventoryWithProductInfo(Long productId) {Inventory inventory = inventoryRepository.findByProductId(productId);// 获取产品基本信息用于库存展示ProductDTO productInfo = productService.getBasicProductInfo(productId);inventory.setProductInfo(productInfo);return inventory;}
}

解决方案：使用Repository模式和DTO组装器

/*** 重构后的解决方案：分离数据访问和业务逻辑*/
@Service
public class ProductQueryService {@Autowiredprivate ProductRepository productRepository;@Autowiredprivate CategoryRepository categoryRepository;@Autowiredprivate InventoryRepository inventoryRepository;@Autowiredprivate PriceRepository priceRepository;@Autowiredprivate ProductDTOAssembler productDTOAssembler;/*** 获取完整的产品信息，避免循环依赖*/public ProductDTO getProductDetails(Long productId) {// 1. 获取基础数据Product product = productRepository.findById(productId);Category category = categoryRepository.findById(product.getCategoryId());Inventory inventory = inventoryRepository.findByProductId(productId);Price price = priceRepository.findByProductId(productId);// 2. 组装DTOreturn productDTOAssembler.assemble(product, category, inventory, price);}
}@Component
public class ProductDTOAssembler {public ProductDTO assemble(Product product, Category category, Inventory inventory, Price price) {return ProductDTO.builder().id(product.getId()).name(product.getName()).description(product.getDescription()).categoryName(category.getName()).categoryPath(category.getPath()).stockQuantity(inventory.getQuantity()).availableQuantity(inventory.getAvailableQuantity()).currentPrice(price.getCurrentPrice()).originalPrice(price.getOriginalPrice()).build();}public CategoryDTO assembleWithProducts(Category category, List<Product> products) {return CategoryDTO.builder().id(category.getId()).name(category.getName()).path(category.getPath()).productCount(products.size()).products(products.stream().map(this::toSimpleProductDTO).collect(Collectors.toList())).build();}private SimpleProductDTO toSimpleProductDTO(Product product) {return SimpleProductDTO.builder().id(product.getId()).name(product.getName()).build();}
}// 专门的查询服务，避免循环依赖
@Service
public class CategoryQueryService {@Autowiredprivate CategoryRepository categoryRepository;@Autowiredprivate ProductRepository productRepository;@Autowiredprivate ProductDTOAssembler assembler;public CategoryDTO getCategoryWithProducts(Long categoryId) {Category category = categoryRepository.findById(categoryId);List<Product> products = productRepository.findByCategoryId(categoryId);return assembler.assembleWithProducts(category, products);}
}

监控和诊断工具

1. 自定义循环依赖监控

@Component
public class CircularDependencyMonitor {private static final Logger logger = LoggerFactory.getLogger(CircularDependencyMonitor.class);private final Map<String, Long> beanCreationTimes = new ConcurrentHashMap<>();private final Map<String, Set<String>> dependencyChains = new ConcurrentHashMap<>();@EventListenerpublic void onBeanCreationStart(BeanCreationStartEvent event) {String beanName = event.getBeanName();beanCreationTimes.put(beanName, System.currentTimeMillis());// 记录依赖链recordDependencyChain(beanName, event.getDependencyChain());}@EventListenerpublic void onBeanCreationEnd(BeanCreationEndEvent event) {String beanName = event.getBeanName();Long startTime = beanCreationTimes.remove(beanName);if (startTime != null) {long duration = System.currentTimeMillis() - startTime;if (duration > 1000) { // 超过1秒的Bean创建logger.warn("Bean {} 创建耗时过长: {}ms, 可能存在循环依赖", beanName, duration);// 分析依赖链analyzeDependencyChain(beanName);}}}private void recordDependencyChain(String beanName, Set<String> chain) {dependencyChains.put(beanName, new HashSet<>(chain));}private void analyzeDependencyChain(String beanName) {Set<String> chain = dependencyChains.get(beanName);if (chain != null && chain.size() > 3) {logger.warn("Bean {} 的依赖链较长: {}", beanName, chain);// 检测是否存在循环if (detectCircularInChain(chain)) {logger.error("检测到循环依赖: {}", chain);}}}private boolean detectCircularInChain(Set<String> chain) {// 简化的循环检测逻辑return chain.size() != new HashSet<>(chain).size();}
}

2. Spring Boot Actuator集成

@Component
@Endpoint(id = "circular-dependencies")
public class CircularDependencyEndpoint {@Autowiredprivate ApplicationContext applicationContext;@ReadOperationpublic Map<String, Object> circularDependencies() {Map<String, Object> result = new HashMap<>();// 分析所有Bean的依赖关系Map<String, Set<String>> dependencyGraph = buildDependencyGraph();// 检测循环依赖List<List<String>> cycles = detectAllCycles(dependencyGraph);result.put("totalBeans", dependencyGraph.size());result.put("circularDependencies", cycles);result.put("affectedBeans", cycles.stream().flatMap(List::stream).distinct().count());return result;}private Map<String, Set<String>> buildDependencyGraph() {Map<String, Set<String>> graph = new HashMap<>();if (applicationContext instanceof ConfigurableApplicationContext) {ConfigurableListableBeanFactory beanFactory = ((ConfigurableApplicationContext) applicationContext).getBeanFactory();String[] beanNames = beanFactory.getBeanDefinitionNames();for (String beanName : beanNames) {try {BeanDefinition bd = beanFactory.getBeanDefinition(beanName);Set<String> dependencies = extractDependencies(bd);graph.put(beanName, dependencies);} catch (Exception e) {// 忽略无法分析的Bean}}}return graph;}private List<List<String>> detectAllCycles(Map<String, Set<String>> graph) {List<List<String>> cycles = new ArrayList<>();Set<String> visited = new HashSet<>();for (String node : graph.keySet()) {if (!visited.contains(node)) {List<String> path = new ArrayList<>();Set<String> pathSet = new HashSet<>();findCycles(graph, node, visited, path, pathSet, cycles);}}return cycles;}private void findCycles(Map<String, Set<String>> graph, String current,Set<String> visited, List<String> path, Set<String> pathSet, List<List<String>> cycles) {if (pathSet.contains(current)) {// 找到循环int cycleStart = path.indexOf(current);List<String> cycle = new ArrayList<>(path.subList(cycleStart, path.size()));cycle.add(current);cycles.add(cycle);return;}if (visited.contains(current)) {return;}visited.add(current);path.add(current);pathSet.add(current);Set<String> neighbors = graph.get(current);if (neighbors != null) {for (String neighbor : neighbors) {findCycles(graph, neighbor, visited, path, pathSet, cycles);}}path.remove(path.size() - 1);pathSet.remove(current);}private Set<String> extractDependencies(BeanDefinition beanDefinition) {// 实现依赖提取逻辑Set<String> dependencies = new HashSet<>();// 提取构造器依赖ConstructorArgumentValues constructorArgs = beanDefinition.getConstructorArgumentValues();extractDependenciesFromConstructorArgs(constructorArgs, dependencies);// 提取属性依赖MutablePropertyValues propertyValues = beanDefinition.getPropertyValues();extractDependenciesFromPropertyValues(propertyValues, dependencies);return dependencies;}
}

3. 性能监控和优化

@Component
public class CircularDependencyPerformanceMonitor {private final MeterRegistry meterRegistry;private final Timer.Sample sample;public CircularDependencyPerformanceMonitor(MeterRegistry meterRegistry) {this.meterRegistry = meterRegistry;this.sample = Timer.start(meterRegistry);}@EventListenerpublic void onApplicationReady(ApplicationReadyEvent event) {sample.stop(Timer.builder("spring.context.startup.time").description("Spring context startup time").register(meterRegistry));// 分析启动过程中的循环依赖处理analyzeStartupPerformance();}private void analyzeStartupPerformance() {ApplicationContext context = applicationContext;// 统计Bean创建耗时Map<String, Long> beanCreationTimes = getBeanCreationTimes();// 识别耗时最长的BeanString slowestBean = beanCreationTimes.entrySet().stream().max(Map.Entry.comparingByValue()).map(Map.Entry::getKey).orElse("unknown");Long slowestTime = beanCreationTimes.get(slowestBean);// 记录指标Gauge.builder("spring.bean.creation.slowest.time").description("Slowest bean creation time").register(meterRegistry, slowestTime);Counter.builder("spring.circular.dependencies.resolved").description("Number of circular dependencies resolved").register(meterRegistry).increment(countResolvedCircularDependencies());}private Map<String, Long> getBeanCreationTimes() {// 实现Bean创建时间统计return new HashMap<>();}private int countResolvedCircularDependencies() {// 统计解决的循环依赖数量return 0;}
}

最佳实践总结

1. 设计原则

/*** 避免循环依赖的设计原则*/
public class CircularDependencyBestPractices {/*** 1. 单一职责原则* 每个类应该只有一个明确的职责，避免过度耦合*/@Servicepublic class UserService {// 只负责用户相关的核心业务逻辑public User createUser(CreateUserRequest request) {// 用户创建逻辑}}/*** 2. 依赖倒置原则* 依赖抽象而不是具体实现*/@Servicepublic class OrderService {// 依赖接口而不是具体实现private final UserRepository userRepository;private final NotificationSender notificationSender;public OrderService(UserRepository userRepository, NotificationSender notificationSender) {this.userRepository = userRepository;this.notificationSender = notificationSender;}}/*** 3. 接口隔离原则* 使用小而专一的接口*/public interface UserValidator {boolean validateUser(User user);}public interface OrderValidator {boolean validateOrder(Order order);}/*** 4. 事件驱动架构* 使用事件解耦服务间的直接依赖*/@Servicepublic class EventDrivenUserService {@Autowiredprivate ApplicationEventPublisher eventPublisher;public User createUser(CreateUserRequest request) {User user = new User(request);user = userRepository.save(user);// 发布事件而不是直接调用其他服务eventPublisher.publishEvent(new UserCreatedEvent(user));return user;}}
}

2. 代码检查清单

/*** 循环依赖检查清单*/
public class CircularDependencyChecklist {/*** 检查项目：* * 1. 构造器注入检查* - 避免构造器循环依赖* - 使用@Lazy注解打破循环* * 2. 服务层设计检查* - 服务职责是否单一* - 是否存在相互调用* * 3. 依赖关系检查* - 依赖层次是否清晰* - 是否违反分层架构原则* * 4. 配置检查* - Bean作用域是否合适* - 是否正确使用@Lazy* * 5. 测试覆盖* - 是否有集成测试覆盖循环依赖场景* - 是否有性能测试验证启动时间*/public void performChecklist() {checkConstructorDependencies();checkServiceLayerDesign();checkDependencyHierarchy();checkConfiguration();checkTestCoverage();}private void checkConstructorDependencies() {// 检查构造器循环依赖}private void checkServiceLayerDesign() {// 检查服务层设计}private void checkDependencyHierarchy() {// 检查依赖层次}private void checkConfiguration() {// 检查配置}private void checkTestCoverage() {// 检查测试覆盖率}
}

结论

Spring通过精妙的三级缓存机制成功解决了单例Bean的循环依赖问题，这一设计体现了Spring框架的深度思考和技术实力。理解这一机制不仅有助于我们更好地使用Spring，还能帮助我们：

关键要点回顾

三级缓存的作用：
- 一级缓存：存储完全初始化的Bean
- 二级缓存：存储早期Bean引用，解决循环依赖
- 三级缓存：存储Bean工厂，处理AOP代理
解决原理：
- 在Bean实例化后立即暴露早期引用
- 通过工厂模式延迟决定是否创建代理
- 确保循环依赖中对象引用的一致性
限制条件：
- 只能解决单例Bean的setter注入循环依赖
- 无法解决构造器循环依赖
- 无法解决prototype作用域的循环依赖
最佳实践：
- 优先使用setter注入而非构造器注入
- 合理使用@Lazy注解
- 采用事件驱动架构解耦服务
- 遵循SOLID设计原则