在某些场景下，会引入熔断机制来保证服务的稳定性。本文介绍用Go语言自己实现断路器需要哪些操作。什么是断路器？熔断是指下游发生错误时，上游主动关闭或限制对下游的请求。原理一般熔断器分为三个时期：CLOSED、OPEN、HALFOPEN。RPC正常时为CLOSED；当RPC错误增加时，熔断器将被触发并进入OPEN；OPEN后经过一定的冷却时间后，熔断器会变为HALFOPEN；HALFOPEN会对下游进行一些战略访问，然后根据结果决定是改为CLOSED还是OPEN；总的来说，三种状态的转换大致如下：去实现https://github.com/rubyist/circuitbreakerIsAllowed是否允许请求，根据当前状态判断CLOSEallowOPEN在CoolingTimeout冷却时间内，冷却时间不允许，状态变为HALFOPEN，在DetectTimeout检测时间内允许访问HALFOPEN，否则不允许访问atomic.StoreInt32((*int32)(&b.state),int32(HALFOPEN))trip判断是否熔断限流达到（可自定义）typeTripFuncfunc(Metricser)boolThresholdTripFunc错误阈值ConsecutiveTripFunc连续错误超过阈值RateTripFunc根据最小访问次数和错误率判断Metricser访问统计，包括成功次数，失败次数，超时次数，错误rate,样本数,连续错误数typeMetricserinterface{Fail()//recordsafailureSucceed()//recordsasuccessTimeout()//recordsatimeoutFailures()int64//返回失败次数Successes()int64//返回成功次数Timeouts()int64//返回超时次数ConseErrors()int64//返回最近的连续错误ErrorRate()float64//rate=(timeouts+failures)/(timeouts+failures+successes)Samples()int64//(timeouts+failures+successes)Counts()(successes,failures,timeoutsint64)Reset()}windowimplementationclasstypewindowstruct{sync.RWMutexoldestint32//oldestbucketindexlatestint32//latestbucketindexbuckets[]bucket//bucketsthiswindowholdsbucketTimetime.Duration//timeeachbucketholdsbucketNumsint32//thenumbeofbucketsinWindowint32//thenumberofbucketsinthewindowallSuccessint64allFailureint64allTimeoutint64conseErrint64}typebucketstruct{failureint64successint64timeoutint64}用环形队列实现动态统计把一个连续的时间切成多个小份，每一个bucket保存BucketTime的统计数据，BucketTime*BucketNums是Statisticaltimeinterval.每一个BucketTime，都会有一个bucketexpired())atomic.AddInt64(&w.allFailure,-oldBucket.Failures())atomic.AddInt64(&w.allTimeout,-oldBucket.Timeouts())w.oldest++ifw.oldest>=w.bucketNums{w.oldest=0}}else{w.inWindow++}w.latest++ifw.latest>=w.bucketNums{w.latest=0}(&w.buckets[w.latest]).Reset()PanelMetricser容器PanelStateChangeHandlerfuseeventtypePanelStateChangeHandlerfunc(keystring,oldState,newStateState,mMetricser)缺陷所有breaker共享同一个BucketTime，统计周期不支持更新冷却时间不支持动态更新
