kubernetes:client-go 系列文章:

  • Kubernetes: client-go 源码剖析(一)
  • Kubernetes: client-go 源码剖析(二)

2.3 运行 informer

运行 informerReflectorinformerindexer 组件关联以实现 informer 流程图的流程。

2.3.1 Reflector List&Watch

运行 informer

informer.Run(stopCh)// client-go/tools/cache/shared_informer.gofunc (s *sharedIndexInformer) Run(stopCh <-chan struct{}) {    func() {...        // 创建 DeltaFIFO 队列fifo := NewDeltaFIFOWithOptions(DeltaFIFOOptions{KnownObjects:          s.indexer,EmitDeltaTypeReplaced: true,Transformer:           s.transform,})cfg := &Config{Queue:             fifo,ListerWatcher:     s.listerWatcher,ObjectType:        s.objectType,ObjectDescription: s.objectDescription,FullResyncPeriod:  s.resyncCheckPeriod,RetryOnError:      false,ShouldResync:      s.processor.shouldResync,Process:           s.HandleDeltas,WatchErrorHandler: s.watchErrorHandler,}        // 根据 Config 创建 informer 的 controllers.controller = New(cfg)s.controller.(*controller).clock = s.clocks.started = true}()    ...    // goroutine 运行 processorwg.StartWithChannel(processorStopCh, s.processor.run)...    // 运行 controllers.controller.Run(stopCh)}

首先,创建队列 Delta FIFO

func (s *sharedIndexInformer) Run(stopCh <-chan struct{}) {...func() {fifo := NewDeltaFIFOWithOptions(DeltaFIFOOptions{KnownObjects:          s.indexer,EmitDeltaTypeReplaced: true,Transformer:           s.transform,})}()}func NewDeltaFIFOWithOptions(opts DeltaFIFOOptions) *DeltaFIFO {...f := &DeltaFIFO{items:        map[string]Deltas{},queue:        []string{},keyFunc:      opts.KeyFunction,knownObjects: opts.KnownObjects,emitDeltaTypeReplaced: opts.EmitDeltaTypeReplaced,transformer:           opts.Transformer,}f.cond.L = &f.lockreturn f}

该队列中存储的是 Delta 资源对象,其存储结构为:

为什么队列要设计成这个样子?因为 informer 在读取队列时,根据 items 的 action type 调用对应 EventHandler 的回调函数。

接下来,实例化 informercontroller 对象,并且调用 controller.Run 运行 controller

func (c *controller) Run(stopCh <-chan struct{}) {...r := NewReflectorWithOptions(c.config.ListerWatcher,c.config.ObjectType,c.config.Queue,ReflectorOptions{ResyncPeriod:    c.config.FullResyncPeriod,TypeDescription: c.config.ObjectDescription,Clock:           c.clock,},)...wg.StartWithChannel(stopCh, r.Run)wait.Until(c.processLoop, time.Second, stopCh)wg.Wait()}

Run 方法中创建 Reflector 核心组件:

func NewReflectorWithOptions(lw ListerWatcher, expectedType interface{}, store Store, options ReflectorOptions) *Reflector {...// Reflector 中包括 ListerWatcher 对象和 DeltaFIFO 队列r := &Reflector{name:            options.Name,resyncPeriod:    options.ResyncPeriod,typeDescription: options.TypeDescription,listerWatcher:   lw,store:           store,...}...return r}

继续进入 wg.StartWithChannel 中运行 Reflector.Run

func (r *Reflector) Run(stopCh <-chan struct{}) {wait.BackoffUntil(func() {// 调用 Reflector 的 ListAndWatch 方法if err := r.ListAndWatch(stopCh); err != nil {r.watchErrorHandler(r, err)}}, r.backoffManager, true, stopCh)}func (r *Reflector) ListAndWatch(stopCh <-chan struct{}) error {...if fallbackToList {err = r.list(stopCh)if err != nil {return err}}...return r.watch(w, stopCh, resyncerrc)}

在这里我们看到 ReflectorListAndWatch 实现了资源的 list 和 watch 操作。这相当于 informer 流程图的第一步。

具体看 Reflectorlist 方法做了什么:

func (r *Reflector) list(stopCh <-chan struct{}) error {...go func() {...pager := pager.New(pager.SimplePageFunc(func(opts metav1.ListOptions) (runtime.Object, error) {return r.listerWatcher.List(opts)}))list, paginatedResult, err = pager.ListWithAlloc(context.Background(), options)}()select {case <-stopCh:return nilcase r := <-panicCh:panic(r)// 阻塞 listcase <-listCh:}...}

首先,在 goroutine 内调用 pager.ListWithAlloc 获得 list 的资源对象:

func (p *ListPager) ListWithAlloc(ctx context.Context, options metav1.ListOptions) (runtime.Object, bool, error) {return p.list(ctx, options, true)}func (p *ListPager) list(ctx context.Context, options metav1.ListOptions, allocNew bool) (runtime.Object, bool, error) {...for {select {case <-ctx.Done():return nil, paginatedResult, ctx.Err()default:}obj, err := p.PageFn(ctx, options)...}...}

list 方法内调用 p.PageFn 获得资源对象 obj,调用 p.PageFn 实际调用的是 Reflector.listerWatcher 对象的 List 方法:

func (lw *ListWatch) List(options metav1.ListOptions) (runtime.Object, error) {return lw.ListFunc(options)}func NewFilteredPodInformer(client kubernetes.Interface, namespace string, resyncPeriod time.Duration, indexers cache.Indexers, tweakListOptions internalinterfaces.TweakListOptionsFunc) cache.SharedIndexInformer {return cache.NewSharedIndexInformer(&cache.ListWatch{ListFunc: func(options metav1.ListOptions) (runtime.Object, error) {if tweakListOptions != nil {tweakListOptions(&options)}return client.CoreV1().Pods(namespace).List(context.TODO(), options)},...})}

可以看到,这里将 Reflector 和前面的 ListFunc 回调函数关联上了,实际通过 ClientSet 客户端对象 list kube-apiserver 的资源。

2.3.2 Reflector Add Object

client-go informer 流程的第一步实现了,那么第二步在哪呢?带着这个问题,我们继续看 Reflector.list 方法:

func (r *Reflector) list(stopCh <-chan struct{}) error {...// 通过反射读取 list 的 meta fieldlistMetaInterface, err := meta.ListAccessor(list)if err != nil {return fmt.Errorf("unable to understand list result %#v: %v", list, err)}// 读取 resource versionresourceVersion = listMetaInterface.GetResourceVersion()items, err := meta.ExtractListWithAlloc(list)if err := r.syncWith(items, resourceVersion); err != nil {return fmt.Errorf("unable to sync list result: %v", err)}return nil}

重点看 Reflector.syncWith 方法:

func (r *Reflector) syncWith(items []runtime.Object, resourceVersion string) error {found := make([]interface{}, 0, len(items))for _, item := range items {found = append(found, item)}return r.store.Replace(found, resourceVersion)}func (f *DeltaFIFO) Replace(list []interface{}, _ string) error {...for _, item := range list {key, err := f.KeyOf(item)if err != nil {return KeyError{item, err}}keys.Insert(key)if err := f.queueActionLocked(action, item); err != nil {return fmt.Errorf("couldn't enqueue object: %v", err)}}...return nil}

Reflector.syncWith 调用 Reflector.DeltaFIFOReplace 方法将 list 的资源对象添加到 DeltaFIFO 队列中,实现 informer 流程的第二步。

watch 资源和 list 资源的流程类似,这里就不过多介绍了。

2.3.3 informer Pop Object

Reflector 作为生产者将 list&watch 的资源添加到 Delta FIFO 队列,那么消费者在哪里使用 Delta FIFO 的资源呢?

client-gocontroller.Run 中的 controller.processLoop 处理 Delta FIFO 的资源:

wait.Until(c.processLoop, time.Second, stopCh)// client-go/tools/cache/controller.gofunc (c *controller) processLoop() {for {obj, err := c.config.Queue.Pop(PopProcessFunc(c.config.Process))if err != nil {if err == ErrFIFOClosed {return}if c.config.RetryOnError {// This is the safe way to re-enqueue.c.config.Queue.AddIfNotPresent(obj)}}}}

controller.processLoop 会轮询队列中的资源,当队列中有资源加入时 Pop 资源:

func (f *DeltaFIFO) Pop(process PopProcessFunc) (interface{}, error) {    for {for len(f.queue) == 0 {...            // 当队列无资源的时候协程阻塞f.cond.Wait()}        id := f.queue[0]f.queue = f.queue[1:]depth := len(f.queue)    ...item, ok := f.items[id]...delete(f.items, id)        ...        err := process(item, isInInitialList)        ...        return item, err    }}

DeltaFIFO.Pop 会循环读取队列中的资源,当队列无资源时进入阻塞状态。如果队列中有资源,每次读取队列的首元素,删除队列中读取的首元素,然后调用回调函数 PopProcessFunc 处理读取的首元素:

err := process(item, isInInitialList)// client-go/tools/cache/shared_informer.gofunc (s *sharedIndexInformer) HandleDeltas(obj interface{}, isInInitialList bool) error {s.blockDeltas.Lock()defer s.blockDeltas.Unlock()if deltas, ok := obj.(Deltas); ok {return processDeltas(s, s.indexer, deltas, isInInitialList)}return errors.New("object given as Process argument is not Deltas")}

调用回调函数 PopProcessFunc 实际调用的是 sharedIndexInformer.HandleDeltas 方法,在该方法内处理从队列读取到的资源。

至此,实现了 informer 流程图的第三步。

2.3.4 informer Add and Store Object

继续看 sharedIndexInformer.HandleDeltas 的函数 processDeltas

func processDeltas(handler ResourceEventHandler, clientState Store, deltas Deltas, isInInitialList bool,) error {for _, d := range deltas {obj := d.Objectswitch d.Type {case Sync, Replaced, Added, Updated:if old, exists, err := clientState.Get(obj); err == nil && exists {if err := clientState.Update(obj); err != nil {return err}handler.OnUpdate(old, obj)} else {if err := clientState.Add(obj); err != nil {return err}handler.OnAdd(obj, isInInitialList)}case Deleted:if err := clientState.Delete(obj); err != nil {return err}handler.OnDelete(obj)}}return nil}

processDeltas 中根据不同的 Delta Type 执行不同的 case。我们以 Added type 为例查看处理流程。

首先,clientState.Get 从本地 indexer 存储中读取资源,并判断资源是否存在:

// clientState.Get get 资源 objfunc (c *cache) Get(obj interface{}) (item interface{}, exists bool, err error) {...return c.GetByKey(key)}func (c *cache) GetByKey(key string) (item interface{}, exists bool, err error) {item, exists = c.cacheStorage.Get(key)return item, exists, nil}// 从 index 中读取资源func (c *threadSafeMap) Get(key string) (item interface{}, exists bool) {c.lock.RLock()defer c.lock.RUnlock()item, exists = c.items[key]return item, exists}

如果资源不存在,进入 cache.Add:

func (c *cache) Add(obj interface{}) error {...    // 添加资源到 indexerc.cacheStorage.Add(key, obj)return nil}func (c *threadSafeMap) Add(key string, obj interface{}) {c.Update(key, obj)}func (c *threadSafeMap) Update(key string, obj interface{}) {c.lock.Lock()defer c.lock.Unlock()oldObject := c.items[key]c.items[key] = objc.index.updateIndices(oldObject, obj, key)}

clientState.Add 中将队列 Delta FIFO 读取的资源存入 indexer 中。

至此,完成了 informer 流程图的第四步和第五步。

2.3.5 Event Handler

将资源存入 indexer 后继续往下看 sharedIndexInformer.OnAdd 是怎么处理 Pop 出的资源的:

func (s *sharedIndexInformer) OnAdd(obj interface{}, isInInitialList bool) {...s.processor.distribute(addNotification{newObj: obj, isInInitialList: isInInitialList}, false)}func (p *sharedProcessor) distribute(obj interface{}, sync bool) {p.listenersLock.RLock()defer p.listenersLock.RUnlock()for listener, isSyncing := range p.listeners {switch {case !sync:// non-sync messages are delivered to every listenerlistener.add(obj)case isSyncing:// sync messages are delivered to every syncing listenerlistener.add(obj)default:// skipping a sync obj for a non-syncing listener}}}func (p *processorListener) add(notification interface{}) {if a, ok := notification.(addNotification); ok && a.isInInitialList {p.syncTracker.Start()}p.addCh <- notification}

可以看到,Pop 的资源被加入 processorListener.addCh 通道。

那么,通道的另一端是哪里在处理呢?

答案在 sharedIndexInformer.Run 中的 sharedProcessor.run:

func (s *sharedIndexInformer) Run(stopCh <-chan struct{}) {    ...    wg.StartWithChannel(processorStopCh, s.processor.run)    ...}func (p *sharedProcessor) run(stopCh <-chan struct{}) {func() {p.listenersLock.RLock()defer p.listenersLock.RUnlock()for listener := range p.listeners {p.wg.Start(listener.run)p.wg.Start(listener.pop)}p.listenersStarted = true}()<-stopCh...}

sharedProcessor.run 方法中开启两个协程分别执行 listener.runlistener.pop 方法。

我们先看 listener.run 方法:

func (p *processorListener) run() {stopCh := make(chan struct{})wait.Until(func() {for next := range p.nextCh {switch notification := next.(type) {case updateNotification:p.handler.OnUpdate(notification.oldObj, notification.newObj)case addNotification:p.handler.OnAdd(notification.newObj, notification.isInInitialList)if notification.isInInitialList {p.syncTracker.Finished()}case deleteNotification:p.handler.OnDelete(notification.oldObj)default:utilruntime.HandleError(fmt.Errorf("unrecognized notification: %T", next))}}close(stopCh)}, 1*time.Second, stopCh)}

listener.runprocessorListener.nextCh 通道中读取资源对象,根据资源对象的类型决定执行哪个 case。

前面通道的一端将 addNotification 加入到 processorListeneraddCh 通道 p.addCh <- notification
这里 processorListener 根据 nextCh 通道的资源执行相应的 case。

那么 addChnextCh 的关联在哪里呢?

我们看 processorListener.pop

func (p *processorListener) pop() {defer utilruntime.HandleCrash()defer close(p.nextCh) // Tell .run() to stopvar nextCh chan<- interface{}var notification interface{}for {select {case nextCh <- notification:// Notification dispatchedvar ok boolnotification, ok = p.pendingNotifications.ReadOne()if !ok { // Nothing to popnextCh = nil // Disable this select case}case notificationToAdd, ok := <-p.addCh:if !ok {return}if notification == nil { // No notification to pop (and pendingNotifications is empty)// Optimize the case - skip adding to pendingNotificationsnotification = notificationToAddnextCh = p.nextCh} else { // There is already a notification waiting to be dispatchedp.pendingNotifications.WriteOne(notificationToAdd)}}}}

processorListener.pop 的逻辑比较复杂,这里不过多介绍。重点在通过 nextCh = p.nextChprocessorListener.nextCh 和函数内通道 nextCh 关联,从而实现 processorListener.addCh 通道到 processorListener.nextCh 通道的数据传递。

了解了通道间的数据传递。我们以 ResourceEventHandlerFuncs.OnAdd 为例看 client-go 是怎么调用 EventHandler 的:

func (p *processorListener) run() {...wait.Until(func() {for next := range p.nextCh {switch notification := next.(type) {case updateNotification:...case addNotification:p.handler.OnAdd(notification.newObj, notification.isInInitialList)...}...}})}func (r ResourceEventHandlerFuncs) OnAdd(obj interface{}, isInInitialList bool) {if r.AddFunc != nil {r.AddFunc(obj)}}func main() {...informer.AddEventHandler(cache.ResourceEventHandlerFuncs{AddFunc: func(obj interface{}) {mObj := obj.(v1.Object)log.Printf("New Pod Added to Store: %s", mObj.GetName())},...})}

可以看到,最终通过回调函数执行我们定义的 AddFunc handler。

至此,实现了 informer 流程图的第六步。

3. 总结

我们通过两篇文章从源码角度介绍了 client-go 的流程。下面要开始 kube-schduler 的学习了,敬请期待…


芝兰生于空谷,不以无人而不芳。