做者:alben
來源:https://albenw.github.io/posts/69a9647f/html
Spring實現了一套重試機制,功能簡單實用。Spring Retry是從Spring Batch獨立出來的一個功能,已經普遍應用於Spring Batch,Spring Integration, Spring for Apache Hadoop等Spring項目。java
本文將講述如何使用Spring Retry及其實現原理。git
重試,其實咱們其實不少時候都須要的,爲了保證容錯性,可用性,一致性等。通常用來應對外部系統的一些不可預料的返回、異常等,特別是網絡延遲,中斷等狀況。還有在如今流行的微服務治理框架中,一般都有本身的重試與超時配置,好比dubbo能夠設置retries=1,timeout=500調用失敗只重試1次,超過500ms調用仍未返回則調用失敗。github
若是咱們要作重試,要爲特定的某個操做作重試功能,則要硬編碼,大概邏輯基本都是寫個循環,根據返回或異常,計數失敗次數,而後設定退出條件。 這樣作,且不說每一個操做都要寫這種相似的代碼,並且重試邏輯和業務邏輯混在一塊兒,給維護和擴展帶來了麻煩。面試
從面向對象的角度來看,咱們應該把重試的代碼獨立出來。spring
先舉個例子:網絡
@Configuration @EnableRetry public class Application { @Bean public RetryService retryService(){ return new RetryService(); } public static void main(String[] args) throws Exception{ ApplicationContext applicationContext = new AnnotationConfigApplicationContext("springretry"); RetryService service1 = applicationContext.getBean("service", RetryService.class); service1.service(); } } @Service("service") public class RetryService { @Retryable(value = IllegalAccessException.class, maxAttempts = 5, backoff= @Backoff(value = 1500, maxDelay = 100000, multiplier = 1.2)) public void service() throws IllegalAccessException { System.out.println("service method..."); throw new IllegalAccessException("manual exception"); } @Recover public void recover(IllegalAccessException e){ System.out.println("service retry after Recover => " + e.getMessage()); } }
@EnableRetry - 表示開啓重試機制intellij-idea
@Retryable - 表示這個方法須要重試,它有很豐富的參數,能夠知足你對重試的需求app
@Backoff - 表示重試中的退避策略框架
@Recover - 兜底方法,即屢次重試後仍是失敗就會執行這個方法
Spring-Retry 的功能豐富在於其重試策略和退避策略,還有兜底,監聽器等操做。
而後每一個註解裏面的參數,都是很簡單的,你們看一下就知道是什麼意思,怎麼用了,我就很少講了。
看一下Spring Retry自帶的一些重試策略,主要是用來判斷當方法調用異常時是否須要重試。(下文原理部分會深刻分析實現)
….等等
看一下退避策略,退避是指怎麼去作下一次的重試,在這裏其實就是等待多長時間。(下文原理部分會深刻分析實現)
原理部分我想分開兩部分來說,一是重試機制的切入點,即它是如何使得你的代碼實現重試功能的;二是重試機制的詳細,包括重試的邏輯以及重試策略和退避策略的實現。另外,關注公衆號Java技術棧,在後臺回覆:面試,能夠獲取我整理的 Spring 系列面試題和答案,很是齊全。
@Target(ElementType.TYPE) @Retention(RetentionPolicy.RUNTIME) @EnableAspectJAutoProxy(proxyTargetClass = false) @Import(RetryConfiguration.class) @Documented public @interface EnableRetry { /** * Indicate whether subclass-based (CGLIB) proxies are to be created as opposed * to standard Java interface-based proxies. The default is {@code false}. * * @return whether to proxy or not to proxy the class */ boolean proxyTargetClass() default false; }
咱們能夠看到
@EnableAspectJAutoProxy(proxyTargetClass = false)
這個並不陌生,就是打開Spring AOP功能。
重點看看@Import(RetryConfiguration.class)
@Import至關於註冊這個Bean
咱們看看這個RetryConfiguration
是個什麼東西:
它是一個AbstractPointcutAdvisor,它有一個pointcut和一個advice。咱們知道,在IOC過程當中會根據PointcutAdvisor類來對Bean進行Pointcut的過濾,而後生成對應的AOP代理類,用advice來增強處理。
看看RetryConfiguration的初始化:
@PostConstruct public void init() { Set<Class<? extends Annotation>> retryableAnnotationTypes = new LinkedHashSet<Class<? extends Annotation>>(1); retryableAnnotationTypes.add(Retryable.class); //建立pointcut this.pointcut = buildPointcut(retryableAnnotationTypes); //建立advice this.advice = buildAdvice(); if (this.advice instanceof BeanFactoryAware) { ((BeanFactoryAware) this.advice).setBeanFactory(beanFactory); } }
protected Pointcut buildPointcut(Set<Class<? extends Annotation>> retryAnnotationTypes) { ComposablePointcut result = null; for (Class<? extends Annotation> retryAnnotationType : retryAnnotationTypes) { Pointcut filter = new AnnotationClassOrMethodPointcut(retryAnnotationType); if (result == null) { result = new ComposablePointcut(filter); } else { result.union(filter); } } return result; }
上面代碼用到了AnnotationClassOrMethodPointcut,其實它最終仍是用到了AnnotationMethodMatcher來根據註解進行切入點的過濾。這裏就是@Retryable註解了。
//建立advice對象,即攔截器 protected Advice buildAdvice() { //下面關注這個對象 AnnotationAwareRetryOperationsInterceptor interceptor = new AnnotationAwareRetryOperationsInterceptor(); if (retryContextCache != null) { interceptor.setRetryContextCache(retryContextCache); } if (retryListeners != null) { interceptor.setListeners(retryListeners); } if (methodArgumentsKeyGenerator != null) { interceptor.setKeyGenerator(methodArgumentsKeyGenerator); } if (newMethodArgumentsIdentifier != null) { interceptor.setNewItemIdentifier(newMethodArgumentsIdentifier); } if (sleeper != null) { interceptor.setSleeper(sleeper); } return interceptor; }
繼承關係
能夠看出AnnotationAwareRetryOperationsInterceptor是一個MethodInterceptor,在建立AOP代理過程當中若是目標方法符合pointcut的規則,它就會加到interceptor列表中,而後作加強,咱們看看invoke方法作了什麼加強。
@Override public Object invoke(MethodInvocation invocation) throws Throwable { MethodInterceptor delegate = getDelegate(invocation.getThis(), invocation.getMethod()); if (delegate != null) { return delegate.invoke(invocation); } else { return invocation.proceed(); } }
這裏用到了委託,主要是須要根據配置委託給具體「有狀態」的interceptor仍是「無狀態」的interceptor。
private MethodInterceptor getDelegate(Object target, Method method) { if (!this.delegates.containsKey(target) || !this.delegates.get(target).containsKey(method)) { synchronized (this.delegates) { if (!this.delegates.containsKey(target)) { this.delegates.put(target, new HashMap<Method, MethodInterceptor>()); } Map<Method, MethodInterceptor> delegatesForTarget = this.delegates.get(target); if (!delegatesForTarget.containsKey(method)) { Retryable retryable = AnnotationUtils.findAnnotation(method, Retryable.class); if (retryable == null) { retryable = AnnotationUtils.findAnnotation(method.getDeclaringClass(), Retryable.class); } if (retryable == null) { retryable = findAnnotationOnTarget(target, method); } if (retryable == null) { return delegatesForTarget.put(method, null); } MethodInterceptor delegate; //支持自定義MethodInterceptor,並且優先級最高 if (StringUtils.hasText(retryable.interceptor())) { delegate = this.beanFactory.getBean(retryable.interceptor(), MethodInterceptor.class); } else if (retryable.stateful()) { //獲得「有狀態」的interceptor delegate = getStatefulInterceptor(target, method, retryable); } else { //獲得「無狀態」的interceptor delegate = getStatelessInterceptor(target, method, retryable); } delegatesForTarget.put(method, delegate); } } } return this.delegates.get(target).get(method); }
getStatefulInterceptor和getStatelessInterceptor都是差很少,咱們先看看比較簡單的getStatelessInterceptor。
private MethodInterceptor getStatelessInterceptor(Object target, Method method, Retryable retryable) { //生成一個RetryTemplate RetryTemplate template = createTemplate(retryable.listeners()); //生成retryPolicy template.setRetryPolicy(getRetryPolicy(retryable)); //生成backoffPolicy template.setBackOffPolicy(getBackoffPolicy(retryable.backoff())); return RetryInterceptorBuilder.stateless() .retryOperations(template) .label(retryable.label()) .recoverer(getRecoverer(target, method)) .build(); }
具體生成retryPolicy和backoffPolicy的規則,咱們等下再回頭來看。
RetryInterceptorBuilder其實就是爲了生成RetryOperationsInterceptor
。RetryOperationsInterceptor也是一個MethodInterceptor,咱們來看看它的invoke
方法。
public Object invoke(final MethodInvocation invocation) throws Throwable { String name; if (StringUtils.hasText(label)) { name = label; } else { name = invocation.getMethod().toGenericString(); } final String label = name; //定義了一個RetryCallback,其實看它的doWithRetry方法,調用了invocation的proceed()方法,是否是有點眼熟,這就是AOP的攔截鏈調用,若是沒有攔截鏈,那就是對原來方法的調用。 RetryCallback<Object, Throwable> retryCallback = new RetryCallback<Object, Throwable>() { public Object doWithRetry(RetryContext context) throws Exception { context.setAttribute(RetryContext.NAME, label); /* * If we don't copy the invocation carefully it won't keep a reference to * the other interceptors in the chain. We don't have a choice here but to * specialise to ReflectiveMethodInvocation (but how often would another * implementation come along?). */ if (invocation instanceof ProxyMethodInvocation) { try { return ((ProxyMethodInvocation) invocation).invocableClone().proceed(); } catch (Exception e) { throw e; } catch (Error e) { throw e; } catch (Throwable e) { throw new IllegalStateException(e); } } else { throw new IllegalStateException( "MethodInvocation of the wrong type detected - this should not happen with Spring AOP, " + "so please raise an issue if you see this exception"); } } }; if (recoverer != null) { ItemRecovererCallback recoveryCallback = new ItemRecovererCallback( invocation.getArguments(), recoverer); return this.retryOperations.execute(retryCallback, recoveryCallback); } //最終仍是進入到retryOperations的execute方法,這個retryOperations就是在以前的builder set進來的RetryTemplate。 return this.retryOperations.execute(retryCallback); }
不管是RetryOperationsInterceptor
仍是StatefulRetryOperationsInterceptor
,最終的攔截處理邏輯仍是調用到RetryTemplate的execute方法,從名字也看出來,RetryTemplate做爲一個模板類,裏面包含了重試統一邏輯。
不過,我看這個RetryTemplate並非很「模板」,由於它沒有不少能夠擴展的地方。
上面介紹了Spring Retry利用了AOP代理使重試機制對業務代碼進行「入侵」。下面咱們繼續看看重試的邏輯作了什麼。
RetryTemplate的doExecute方法。
protected <T, E extends Throwable> T doExecute(RetryCallback<T, E> retryCallback, RecoveryCallback<T> recoveryCallback, RetryState state) throws E, ExhaustedRetryException { RetryPolicy retryPolicy = this.retryPolicy; BackOffPolicy backOffPolicy = this.backOffPolicy; //新建一個RetryContext來保存本輪重試的上下文 RetryContext context = open(retryPolicy, state); if (this.logger.isTraceEnabled()) { this.logger.trace("RetryContext retrieved: " + context); } // Make sure the context is available globally for clients who need // it... RetrySynchronizationManager.register(context); Throwable lastException = null; boolean exhausted = false; try { //若是有註冊RetryListener,則會調用它的open方法,給調用者一個通知。 boolean running = doOpenInterceptors(retryCallback, context); if (!running) { throw new TerminatedRetryException( "Retry terminated abnormally by interceptor before first attempt"); } // Get or Start the backoff context... BackOffContext backOffContext = null; Object resource = context.getAttribute("backOffContext"); if (resource instanceof BackOffContext) { backOffContext = (BackOffContext) resource; } if (backOffContext == null) { backOffContext = backOffPolicy.start(context); if (backOffContext != null) { context.setAttribute("backOffContext", backOffContext); } } //判斷可否重試,就是調用RetryPolicy的canRetry方法來判斷。 //這個循環會直到原方法不拋出異常,或不須要再重試 while (canRetry(retryPolicy, context) && !context.isExhaustedOnly()) { try { if (this.logger.isDebugEnabled()) { this.logger.debug("Retry: count=" + context.getRetryCount()); } //清除上次記錄的異常 lastException = null; //doWithRetry方法,通常來講就是原方法 return retryCallback.doWithRetry(context); } catch (Throwable e) { //原方法拋出了異常 lastException = e; try { //記錄異常信息 registerThrowable(retryPolicy, state, context, e); } catch (Exception ex) { throw new TerminatedRetryException("Could not register throwable", ex); } finally { //調用RetryListener的onError方法 doOnErrorInterceptors(retryCallback, context, e); } //再次判斷可否重試 if (canRetry(retryPolicy, context) && !context.isExhaustedOnly()) { try { //若是能夠重試則走退避策略 backOffPolicy.backOff(backOffContext); } catch (BackOffInterruptedException ex) { lastException = e; // back off was prevented by another thread - fail the retry if (this.logger.isDebugEnabled()) { this.logger .debug("Abort retry because interrupted: count=" + context.getRetryCount()); } throw ex; } } if (this.logger.isDebugEnabled()) { this.logger.debug( "Checking for rethrow: count=" + context.getRetryCount()); } if (shouldRethrow(retryPolicy, context, state)) { if (this.logger.isDebugEnabled()) { this.logger.debug("Rethrow in retry for policy: count=" + context.getRetryCount()); } throw RetryTemplate.<E>wrapIfNecessary(e); } } /* * A stateful attempt that can retry may rethrow the exception before now, * but if we get this far in a stateful retry there's a reason for it, * like a circuit breaker or a rollback classifier. */ if (state != null && context.hasAttribute(GLOBAL_STATE)) { break; } } if (state == null && this.logger.isDebugEnabled()) { this.logger.debug( "Retry failed last attempt: count=" + context.getRetryCount()); } exhausted = true; //重試結束後若是有兜底Recovery方法則執行,不然拋異常 return handleRetryExhausted(recoveryCallback, context, state); } catch (Throwable e) { throw RetryTemplate.<E>wrapIfNecessary(e); } finally { //處理一些關閉邏輯 close(retryPolicy, context, state, lastException == null || exhausted); //調用RetryListener的close方法 doCloseInterceptors(retryCallback, context, lastException); RetrySynchronizationManager.clear(); } }
主要核心重試邏輯就是上面的代碼了,看上去仍是挺簡單的。
在上面,咱們漏掉了RetryPolicy的canRetry方法和BackOffPolicy的backOff方法,以及這兩個Policy是怎麼來的。
咱們回頭看看getStatelessInterceptor
方法中的getRetryPolicy
和getRetryPolicy
方法。
private RetryPolicy getRetryPolicy(Annotation retryable) { Map<String, Object> attrs = AnnotationUtils.getAnnotationAttributes(retryable); @SuppressWarnings("unchecked") Class<? extends Throwable>[] includes = (Class<? extends Throwable>[]) attrs.get("value"); String exceptionExpression = (String) attrs.get("exceptionExpression"); boolean hasExpression = StringUtils.hasText(exceptionExpression); if (includes.length == 0) { @SuppressWarnings("unchecked") Class<? extends Throwable>[] value = (Class<? extends Throwable>[]) attrs.get("include"); includes = value; } @SuppressWarnings("unchecked") Class<? extends Throwable>[] excludes = (Class<? extends Throwable>[]) attrs.get("exclude"); Integer maxAttempts = (Integer) attrs.get("maxAttempts"); String maxAttemptsExpression = (String) attrs.get("maxAttemptsExpression"); if (StringUtils.hasText(maxAttemptsExpression)) { maxAttempts = PARSER.parseExpression(resolve(maxAttemptsExpression), PARSER_CONTEXT) .getValue(this.evaluationContext, Integer.class); } if (includes.length == 0 && excludes.length == 0) { SimpleRetryPolicy simple = hasExpression ? new ExpressionRetryPolicy(resolve(exceptionExpression)) .withBeanFactory(this.beanFactory) : new SimpleRetryPolicy(); simple.setMaxAttempts(maxAttempts); return simple; } Map<Class<? extends Throwable>, Boolean> policyMap = new HashMap<Class<? extends Throwable>, Boolean>(); for (Class<? extends Throwable> type : includes) { policyMap.put(type, true); } for (Class<? extends Throwable> type : excludes) { policyMap.put(type, false); } boolean retryNotExcluded = includes.length == 0; if (hasExpression) { return new ExpressionRetryPolicy(maxAttempts, policyMap, true, exceptionExpression, retryNotExcluded) .withBeanFactory(this.beanFactory); } else { return new SimpleRetryPolicy(maxAttempts, policyMap, true, retryNotExcluded); } }
嗯~,代碼不難,這裏簡單作一下總結好了。就是經過@Retryable註解中的參數,來判斷具體使用文章開頭說到的哪一個重試策略,是SimpleRetryPolicy仍是ExpressionRetryPolicy等。
private BackOffPolicy getBackoffPolicy(Backoff backoff) { long min = backoff.delay() == 0 ? backoff.value() : backoff.delay(); if (StringUtils.hasText(backoff.delayExpression())) { min = PARSER.parseExpression(resolve(backoff.delayExpression()), PARSER_CONTEXT) .getValue(this.evaluationContext, Long.class); } long max = backoff.maxDelay(); if (StringUtils.hasText(backoff.maxDelayExpression())) { max = PARSER.parseExpression(resolve(backoff.maxDelayExpression()), PARSER_CONTEXT) .getValue(this.evaluationContext, Long.class); } double multiplier = backoff.multiplier(); if (StringUtils.hasText(backoff.multiplierExpression())) { multiplier = PARSER.parseExpression(resolve(backoff.multiplierExpression()), PARSER_CONTEXT) .getValue(this.evaluationContext, Double.class); } if (multiplier > 0) { ExponentialBackOffPolicy policy = new ExponentialBackOffPolicy(); if (backoff.random()) { policy = new ExponentialRandomBackOffPolicy(); } policy.setInitialInterval(min); policy.setMultiplier(multiplier); policy.setMaxInterval(max > min ? max : ExponentialBackOffPolicy.DEFAULT_MAX_INTERVAL); if (this.sleeper != null) { policy.setSleeper(this.sleeper); } return policy; } if (max > min) { UniformRandomBackOffPolicy policy = new UniformRandomBackOffPolicy(); policy.setMinBackOffPeriod(min); policy.setMaxBackOffPeriod(max); if (this.sleeper != null) { policy.setSleeper(this.sleeper); } return policy; } FixedBackOffPolicy policy = new FixedBackOffPolicy(); policy.setBackOffPeriod(min); if (this.sleeper != null) { policy.setSleeper(this.sleeper); } return policy; }
嗯~,同樣的味道。就是經過@Backoff註解中的參數,來判斷具體使用文章開頭說到的哪一個退避策略,是FixedBackOffPolicy仍是UniformRandomBackOffPolicy等。
那麼每一個RetryPolicy都會重寫canRetry方法,而後在RetryTemplate判斷是否須要重試。
咱們看看SimpleRetryPolicy的
@Override public boolean canRetry(RetryContext context) { Throwable t = context.getLastThrowable(); //判斷拋出的異常是否符合重試的異常 //還有,是否超過了重試的次數 return (t == null || retryForException(t)) && context.getRetryCount() < maxAttempts; }
一樣,咱們看看FixedBackOffPolicy的退避方法。
protected void doBackOff() throws BackOffInterruptedException { try { //就是sleep固定的時間 sleeper.sleep(backOffPeriod); } catch (InterruptedException e) { throw new BackOffInterruptedException("Thread interrupted while sleeping", e); } }
至此,重試的主要原理以及邏輯大概就是這樣了。
我以爲有必要說說RetryContext,先看看它的繼承關係。
能夠看出對每個策略都有對應的Context。
在Spring Retry裏,其實每個策略都是單例來的。我剛開始直覺是對每個須要重試的方法都會new一個策略,這樣重試策略之間纔不會產生衝突,可是一想就知道這樣就可能多出了不少策略對象出來,增長了使用者的負擔,這不是一個好的設計。
Spring Retry採用了一個更加輕量級的作法,就是針對每個須要重試的方法只new一個上下文Context對象,而後在重試時,把這個Context傳到策略裏,策略再根據這個Context作重試,並且Spring Retry還對這個Context作了cache。這樣就至關於對重試的上下文作了優化。
Spring Retry經過AOP機制來實現對業務代碼的重試」入侵「,RetryTemplate中包含了核心的重試邏輯,還提供了豐富的重試策略和退避策略。另外,關注公衆號Java技術棧,在後臺回覆:面試,能夠獲取我整理的 Spring 系列面試題和答案,很是齊全。
參考資料:
http://www.10tiao.com/html/164/201705/2652898434/1.html
https://www.jianshu.com/p/58e753ca0151
https://paper.tuisec.win/detail/90bd660fad92183
近期熱文推薦:
1.600+ 道 Java面試題及答案整理(2021最新版)
2.終於靠開源項目弄到 IntelliJ IDEA 激活碼了,真香!
3.阿里 Mock 工具正式開源,幹掉市面上全部 Mock 工具!
4.Spring Cloud 2020.0.0 正式發佈,全新顛覆性版本!
以爲不錯,別忘了隨手點贊+轉發哦!