想系统学习k8s源码，云原生的可以加：mkjnnm
Informer 通过对 APIServer 的资源对象执行 List 和 Watch 操作，把获取到的数据存储在本地的缓存中，其中实现这个的核心功能就是 Reflector，我们可以称其为反射器，从名字我们可以看出来它的主要功能就是反射，就是将 Etcd 里面的数据(Api Server的数据)反射到本地存储（DeltaFIFO）中。
本篇文章基于 k8s源码v1.30.0讲解
结构体定义
// Reflector watches a specified resource and causes all changes to be reflected in the given store.
type Reflector struct {
// name 标识这个反射器的名称，默认为 文件:行数（比如reflector.go:125）
name string
// The name of the type we expect to place in the store. The name
// will be the stringification of expectedGVK if provided, and the
// stringification of expectedType otherwise. It is for display
// only, and should not be used for parsing or comparison.
typeDescription string
// An example object of the type we expect to place in the store.
// Only the type needs to be right, except that when that is
// `unstructured.Unstructured` the object's `"apiVersion"` and
// `"kind"` must also be right.
expectedType reflect.Type
// The GVK of the object we expect to place in the store if unstructured.
expectedGVK *schema.GroupVersionKind
// The destination to sync up with the watch source
store Store
// listerWatcher is used to perform lists and watches.
listerWatcher ListerWatcher
// backoff manages backoff of ListWatch
backoffManager wait.BackoffManager
resyncPeriod time.Duration
// minWatchTimeout defines the minimum timeout for watch requests.
minWatchTimeout time.Duration
// clock allows tests to manipulate time
clock clock.Clock
// paginatedResult defines whether pagination should be forced for list calls.
// It is set based on the result of the initial list call.
paginatedResult bool
// lastSyncResourceVersion is the resource version token last
// observed when doing a sync with the underlying store
// it is thread safe, but not synchronized with the underlying store
lastSyncResourceVersion string
// isLastSyncResourceVersionUnavailable is true if the previous list or watch request with
// lastSyncResourceVersion failed with an "expired" or "too large resource version" error.
isLastSyncResourceVersionUnavailable bool
// lastSyncResourceVersionMutex guards read/write access to lastSyncResourceVersion
lastSyncResourceVersionMutex sync.RWMutex
// Called whenever the ListAndWatch drops the connection with an error.
watchErrorHandler WatchErrorHandler
// WatchListPageSize is the requested chunk size of initial and resync watch lists.
// If unset, for consistent reads (RV="") or reads that opt-into arbitrarily old data
// (RV="0") it will default to pager.PageSize, for the rest (RV != "" && RV != "0")
// it will turn off pagination to allow serving them from watch cache.
// NOTE: It should be used carefully as paginated lists are always served directly from
// etcd, which is significantly less efficient and may lead to serious performance and
// scalability problems.
WatchListPageSize int64
// ShouldResync is invoked periodically and whenever it returns `true` the Store's Resync operation is invoked
ShouldResync func() bool
// MaxInternalErrorRetryDuration defines how long we should retry internal errors returned by watch.
MaxInternalErrorRetryDuration time.Duration
// UseWatchList if turned on instructs the reflector to open a stream to bring data from the API server.
// Streaming has the primary advantage of using fewer server's resources to fetch data.
//
// The old behaviour establishes a LIST request which gets data in chunks.
// Paginated list is less efficient and depending on the actual size of objects
// might result in an increased memory consumption of the APIServer.
//
// See https://github.com/kubernetes/enhancements/tree/master/keps/sig-api-machinery/3157-watch-list#design-details
//
// TODO(#115478): Consider making reflector.UseWatchList a private field. Since we implemented "api streaming" on the etcd storage layer it should work.
UseWatchList *bool
}
初始化
通过 NewReflector 实例化反射器的时候，必须传入一个 ListerWatcher 接口对象，这个也是反射器最核心的功能，该接口拥有 List 和 Watch 方法，用于获取和监控资源对象。
// NewReflector creates a new Reflector with its name defaulted to the closest source_file.go:line in the call stack
// that is outside this package. See NewReflectorWithOptions for further information.
func NewReflector(lw ListerWatcher, expectedType interface{}, store Store, resyncPeriod time.Duration) *Reflector {
return NewReflectorWithOptions(lw, expectedType, store, ReflectorOptions{ResyncPeriod: resyncPeriod})
}
// NewReflectorWithOptions creates a new Reflector object which will keep the
// given store up to date with the server's contents for the given
// resource. Reflector promises to only put things in the store that
// have the type of expectedType, unless expectedType is nil. If
// resyncPeriod is non-zero, then the reflector will periodically
// consult its ShouldResync function to determine whether to invoke
// the Store's Resync operation; `ShouldResync==nil` means always
// "yes". This enables you to use reflectors to periodically process
// everything as well as incrementally processing the things that
// change.
func NewReflectorWithOptions(lw ListerWatcher, expectedType interface{}, store Store, options ReflectorOptions) *Reflector {
reflectorClock := options.Clock
if reflectorClock == nil {
reflectorClock = clock.RealClock{}
}
minWatchTimeout := defaultMinWatchTimeout
if options.MinWatchTimeout > defaultMinWatchTimeout {
minWatchTimeout = options.MinWatchTimeout
}
r := &Reflector{
name: options.Name,
resyncPeriod: options.ResyncPeriod,
minWatchTimeout: minWatchTimeout,
typeDescription: options.TypeDescription,
listerWatcher: lw,
store: store,
// We used to make the call every 1sec (1 QPS), the goal here is to achieve ~98% traffic reduction when
// API server is not healthy. With these parameters, backoff will stop at [30,60) sec interval which is
// 0.22 QPS. If we don't backoff for 2min, assume API server is healthy and we reset the backoff.
backoffManager: wait.NewExponentialBackoffManager(800*time.Millisecond, 30*time.Second, 2*time.Minute, 2.0, 1.0, reflectorClock),
clock: reflectorClock,
watchErrorHandler: WatchErrorHandler(DefaultWatchErrorHandler),
expectedType: reflect.TypeOf(expectedType),
}
if r.name == "" {
r.name = naming.GetNameFromCallsite(internalPackages...)
}
if r.typeDescription == "" {
r.typeDescription = getTypeDescriptionFromObject(expectedType)
}
if r.expectedGVK == nil {
r.expectedGVK = getExpectedGVKFromObject(expectedType)
}
// don't overwrite UseWatchList if already set
// because the higher layers (e.g. storage/cacher) disabled it on purpose
if r.UseWatchList == nil {
if s := os.Getenv("ENABLE_CLIENT_GO_WATCH_LIST_ALPHA"); len(s) > 0 {
r.UseWatchList = ptr.To(true)
}
}
return r
}
重点需要关注的参数：
-
ListerWatcher ： 每个资源都需要实现的2种动作，用来从apiserver同步数据 -
expectedType： 期望放到 Store 中的类型，如果是unstructured.Unstructured，其`“apiVersion”`和“Kind”也必须是正确的。 -
Store： Deltafifo 的一个实例
启动
启动 refactor 就是调用 Run方法，Run方法会调动 ListAndWatch 方法
逻辑大体分为三大块：
A.List操作（只执行一次）：
（1）设置ListOptions，将ResourceVersion设置为“0”；
（2）调用r.listerWatcher.List方法，执行list操作，即获取全量的资源对象；
（3）根据list回来的资源对象，获取最新的resourceVersion；
（4）资源转换，将list操作获取回来的结果转换为[]runtime.Object结构；
（5）调用r.syncWith，根据list回来转换后的结果去替换store里的items；
（6）调用r.setLastSyncResourceVersion，为Reflector更新已被处理的最新资源对象的resourceVersion值；
B.Resync操作（异步循环执行）；
（1）判断是否需要执行Resync操作，即重新同步；
（2）需要则调用r.store.Resync操作后端store做处理；
C.Watch操作（循环执行）：
（1）stopCh处理，判断是否需要退出循环；
（2）设置ListOptions，设置resourceVersion为最新的resourceVersion，即从list回来的最新resourceVersion开始执行watch操作；
（3）调用r.listerWatcher.Watch，开始监听操作；
（4）watch监听操作的错误返回处理；
（5）调用r.watchHandler，处理watch操作返回来的结果，操作后端store，新增、更新或删除items；
func (r *Reflector) Run(stopCh <-chan struct{}) {
klog.V(3).Infof("Starting reflector %s (%s) from %s", r.typeDescription, r.resyncPeriod, r.name)
wait.BackoffUntil(func() {
if err := r.ListAndWatch(stopCh); err != nil {
r.watchErrorHandler(r, err)
}
}, r.backoffManager, true, stopCh)
klog.V(3).Infof("Stopping reflector %s (%s) from %s", r.typeDescription, r.resyncPeriod, r.name)
}
// ListAndWatch first lists all items and get the resource version at the moment of call,
// and then use the resource version to watch.
// It returns error if ListAndWatch didn't even try to initialize watch.
func (r *Reflector) ListAndWatch(stopCh <-chan struct{}) error {
klog.V(3).Infof("Listing and watching %v from %s", r.typeDescription, r.name)
var err error
var w watch.Interface
useWatchList := ptr.Deref(r.UseWatchList, false)
fallbackToList := !useWatchList
if useWatchList {
w, err = r.watchList(stopCh)
if w == nil && err == nil {
// stopCh was closed
return nil
}
if err != nil {
klog.Warningf("The watchlist request ended with an error, falling back to the standard LIST/WATCH semantics because making progress is better than deadlocking, err = %v", err)
fallbackToList = true
// ensure that we won't accidentally pass some garbage down the watch.
w = nil
}
}
if fallbackToList {
err = r.list(stopCh)
if err != nil {
return err
}
}
klog.V(2).Infof("Caches populated for %v from %s", r.typeDescription, r.name)
resyncerrc := make(chan error, 1)
cancelCh := make(chan struct{})
defer close(cancelCh)
go r.startResync(stopCh, cancelCh, resyncerrc)
return r.watch(w, stopCh, resyncerrc)
}
A.List操作（只执行一次）：
// list simply lists all items and records a resource version obtained from the server at the moment of the call.
// the resource version can be used for further progress notification (aka. watch).
func (r *Reflector) list(stopCh <-chan struct{}) error {
var resourceVersion string
options := metav1.ListOptions{ResourceVersion: r.relistResourceVersion()}
initTrace := trace.New("Reflector ListAndWatch", trace.Field{Key: "name", Value: r.name})
defer initTrace.LogIfLong(10 * time.Second)
var list runtime.Object
var paginatedResult bool
var err error
listCh := make(chan struct{}, 1)
panicCh := make(chan interface{}, 1)
go func() {
defer func() {
if r := recover(); r != nil {
panicCh <- r
}
}()
// Attempt to gather list in chunks, if supported by listerWatcher, if not, the first
// list request will return the full response.
pager := pager.New(pager.SimplePageFunc(func(opts metav1.ListOptions) (runtime.Object, error) {
return r.listerWatcher.List(opts)
}))
switch {
case r.WatchListPageSize != 0:
pager.PageSize = r.WatchListPageSize
case r.paginatedResult:
// We got a paginated result initially. Assume this resource and server honor
// paging requests (i.e. watch cache is probably disabled) and leave the default
// pager size set.
case options.ResourceVersion != "" && options.ResourceVersion != "0":
// User didn't explicitly request pagination.
//
// With ResourceVersion != "", we have a possibility to list from watch cache,
// but we do that (for ResourceVersion != "0") only if Limit is unset.
// To avoid thundering herd on etcd (e.g. on master upgrades), we explicitly
// switch off pagination to force listing from watch cache (if enabled).
// With the existing semantic of RV (result is at least as fresh as provided RV),
// this is correct and doesn't lead to going back in time.
//
// We also don't turn off pagination for ResourceVersion="0", since watch cache
// is ignoring Limit in that case anyway, and if watch cache is not enabled
// we don't introduce regression.
pager.PageSize = 0
}
list, paginatedResult, err = pager.ListWithAlloc(context.Background(), options)
if isExpiredError(err) || isTooLargeResourceVersionError(err) {
r.setIsLastSyncResourceVersionUnavailable(true)
// Retry immediately if the resource version used to list is unavailable.
// The pager already falls back to full list if paginated list calls fail due to an "Expired" error on
// continuation pages, but the pager might not be enabled, the full list might fail because the
// resource version it is listing at is expired or the cache may not yet be synced to the provided
// resource version. So we need to fallback to resourceVersion="" in all to recover and ensure
// the reflector makes forward progress.
list, paginatedResult, err = pager.ListWithAlloc(context.Background(), metav1.ListOptions{ResourceVersion: r.relistResourceVersion()})
}
close(listCh)
}()
select {
case <-stopCh:
return nil
case r := <-panicCh:
panic(r)
case <-listCh:
}
initTrace.Step("Objects listed", trace.Field{Key: "error", Value: err})
if err != nil {
klog.Warningf("%s: failed to list %v: %v", r.name, r.typeDescription, err)
return fmt.Errorf("failed to list %v: %w", r.typeDescription, err)
}
// We check if the list was paginated and if so set the paginatedResult based on that.
// However, we want to do that only for the initial list (which is the only case
// when we set ResourceVersion="0"). The reasoning behind it is that later, in some
// situations we may force listing directly from etcd (by setting ResourceVersion="")
// which will return paginated result, even if watch cache is enabled. However, in
// that case, we still want to prefer sending requests to watch cache if possible.
//
// Paginated result returned for request with ResourceVersion="0" mean that watch
// cache is disabled and there are a lot of objects of a given type. In such case,
// there is no need to prefer listing from watch cache.
if options.ResourceVersion == "0" && paginatedResult {
r.paginatedResult = true
}
r.setIsLastSyncResourceVersionUnavailable(false) // list was successful
listMetaInterface, err := meta.ListAccessor(list)
if err != nil {
return fmt.Errorf("unable to understand list result %#v: %v", list, err)
}
resourceVersion = listMetaInterface.GetResourceVersion()
initTrace.Step("Resource version extracted")
items, err := meta.ExtractListWithAlloc(list)
if err != nil {
return fmt.Errorf("unable to understand list result %#v (%v)", list, err)
}
initTrace.Step("Objects extracted")
if err := r.syncWith(items, resourceVersion); err != nil {
return fmt.Errorf("unable to sync list result: %v", err)
}
initTrace.Step("SyncWith done")
r.setLastSyncResourceVersion(resourceVersion)
initTrace.Step("Resource version updated")
return nil
B.Resync操作（异步循环执行）；
// startResync periodically calls r.store.Resync() method.
// Note that this method is blocking and should be
// called in a separate goroutine.
func (r *Reflector) startResync(stopCh <-chan struct{}, cancelCh <-chan struct{}, resyncerrc chan error) {
resyncCh, cleanup := r.resyncChan()
defer func() {
cleanup() // Call the last one written into cleanup
}()
for {
select {
case <-resyncCh:
case <-stopCh:
return
case <-cancelCh:
return
}
if r.ShouldResync == nil || r.ShouldResync() {
klog.V(4).Infof("%s: forcing resync", r.name)
if err := r.store.Resync(); err != nil {
resyncerrc <- err
return
}
}
cleanup()
resyncCh, cleanup = r.resyncChan()
}
}
C.Watch操作（循环执行）：
// watch simply starts a watch request with the server.
func (r *Reflector) watch(w watch.Interface, stopCh <-chan struct{}, resyncerrc chan error) error {
var err error
retry := NewRetryWithDeadline(r.MaxInternalErrorRetryDuration, time.Minute, apierrors.IsInternalError, r.clock)
for {
// give the stopCh a chance to stop the loop, even in case of continue statements further down on errors
select {
case <-stopCh:
// we can only end up here when the stopCh
// was closed after a successful watchlist or list request
if w != nil {
w.Stop()
}
return nil
default:
}
// start the clock before sending the request, since some proxies won't flush headers until after the first watch event is sent
start := r.clock.Now()
if w == nil {
timeoutSeconds := int64(r.minWatchTimeout.Seconds() * (rand.Float64() + 1.0))
options := metav1.ListOptions{
ResourceVersion: r.LastSyncResourceVersion(),
// We want to avoid situations of hanging watchers. Stop any watchers that do not
// receive any events within the timeout window.
TimeoutSeconds: &timeoutSeconds,
// To reduce load on kube-apiserver on watch restarts, you may enable watch bookmarks.
// Reflector doesn't assume bookmarks are returned at all (if the server do not support
// watch bookmarks, it will ignore this field).
AllowWatchBookmarks: true,
}
w, err = r.listerWatcher.Watch(options)
if err != nil {
if canRetry := isWatchErrorRetriable(err); canRetry {
klog.V(4).Infof("%s: watch of %v returned %v - backing off", r.name, r.typeDescription, err)
select {
case <-stopCh:
return nil
case <-r.backoffManager.Backoff().C():
continue
}
}
return err
}
}
err = watchHandler(start, w, r.store, r.expectedType, r.expectedGVK, r.name, r.typeDescription, r.setLastSyncResourceVersion, nil, r.clock, resyncerrc, stopCh)
// Ensure that watch will not be reused across iterations.
w.Stop()
w = nil
retry.After(err)
if err != nil {
if err != errorStopRequested {
switch {
case isExpiredError(err):
// Don't set LastSyncResourceVersionUnavailable - LIST call with ResourceVersion=RV already
// has a semantic that it returns data at least as fresh as provided RV.
// So first try to LIST with setting RV to resource version of last observed object.
klog.V(4).Infof("%s: watch of %v closed with: %v", r.name, r.typeDescription, err)
case apierrors.IsTooManyRequests(err):
klog.V(2).Infof("%s: watch of %v returned 429 - backing off", r.name, r.typeDescription)
select {
case <-stopCh:
return nil
case <-r.backoffManager.Backoff().C():
continue
}
case apierrors.IsInternalError(err) && retry.ShouldRetry():
klog.V(2).Infof("%s: retrying watch of %v internal error: %v", r.name, r.typeDescription, err)
continue
default:
klog.Warningf("%s: watch of %v ended with: %v", r.name, r.typeDescription, err)
}
}
return nil
}
}
1512
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
