kubernetes endpoint 代码阅读

打个广告，如果想旅游（国内外周边，签证，酒店，门票等）可以联系 15713688208 ，微信同号。

常量参数

首先看一些常量参数：

maxRetries: 服务在退出队列之前的重试次数。会因为限速控制器，重试间隔越来越长。

maxCapacity: 存储在endpoint资源中的最大地址数，在未来的版本中可能会超过此长度的控制。

const (
  // maxRetries is the number of times a service will be retried before it is dropped out of the queue.
  // With the current rate-limiter in use (5ms*2^(maxRetries-1)) the following numbers represent the
  // sequence of delays between successive queuings of a service.
  //
  // 5ms, 10ms, 20ms, 40ms, 80ms, 160ms, 320ms, 640ms, 1.3s, 2.6s, 5.1s, 10.2s, 20.4s, 41s, 82s
  maxRetries = 15

  // maxCapacity represents the maximum number of addresses that should be
  // stored in an Endpoints resource. In a future release, this controller
  // may truncate endpoints exceeding this length.
  maxCapacity = 1000

  // TolerateUnreadyEndpointsAnnotation is an annotation on the Service denoting if the endpoints
  // controller should go ahead and create endpoints for unready pods. This annotation is
  // currently only used by StatefulSets, where we need the pod to be DNS
  // resolvable during initialization and termination. In this situation we
  // create a headless Service just for the StatefulSet, and clients shouldn't
  // be using this Service for anything so unready endpoints don't matter.
  // Endpoints of these Services retain their DNS records and continue
  // receiving traffic for the Service from the moment the kubelet starts all
  // containers in the pod and marks it "Running", till the kubelet stops all
  // containers and deletes the pod from the apiserver.
  // This field is deprecated. v1.Service.PublishNotReadyAddresses will replace it
  // subsequent releases.  It will be removed no sooner than 1.13.
  TolerateUnreadyEndpointsAnnotation = "service.alpha.kubernetes.io/tolerate-unready-endpoints"
)

执行流程

kube-controller-manager 中可以设置参数，在同一时间处理的endpoint数量，可以调整的更大，那么相应的就会耗费更加多的cpu和负载。

–concurrent-endpoint-syncs int32 Default: 5

The number of endpoint syncing operations that will be done concurrently. Larger number = faster endpoint updating, but more CPU (and network) load

用于批量pod更新，默认为0值，表示每个pod都会触发更新。

–endpoint-updates-batch-period duration

The length of endpoint updates batching period. Processing of pod changes will be delayed by this duration to join them with potential upcoming updates and reduce the overall number of endpoints updates. Larger number = higher endpoint programming latency, but lower number of endpoints revision generated

// EndpointControllerConfiguration contains elements describing EndpointController.
type EndpointControllerConfiguration struct {
// concurrentEndpointSyncs is the number of endpoint syncing operations
// that will be done concurrently. Larger number = faster endpoint updating,
// but more CPU (and network) load.
ConcurrentEndpointSyncs int32


// EndpointUpdatesBatchPeriod can be used to batch endpoint updates.
// All updates of endpoint triggered by pod change will be delayed by up to
// 'EndpointUpdatesBatchPeriod'. If other pods in the same endpoint change
// in that period, they will be batched to a single endpoint update.
// Default 0 value means that each pod update triggers an endpoint update.
EndpointUpdatesBatchPeriod metav1.Duration
}

开始执行endpoint进行调用，传入响应的参数。

func startEndpointController(ctx ControllerContext) (http.Handler, bool, error) {
  go endpointcontroller.NewEndpointController(
    ctx.InformerFactory.Core().V1().Pods(),