专注于大数据及容器云核心技术解密，可提供全栈的大数据+云原生平台咨询方案，请持续关注本套博客。如有任何学术交流，可随时联系。更多内容请关注《数据云技术社区》公众号。
1 kubelet 代码目录
- kubernetes/cmd/kubelet部分主要对运行参数进行定义和解析，初始化和构造相关的依赖组件（主要在kubeDeps结构体中）。kubelet运行的详细逻辑，该部分位于kubernetes/pkg/kubelet模块。
- kubernetes 是一个分布式的集群管理系统，在每个节点（node）上都要运行一个 worker 对容器进行生命周期的管理，这个 worker 程序就是 kubelet。
- kubelet 的主要功能就是定时从某个地方获取节点上 pod/container 的期望状态（运行什么容器、运行的副本数量、网络或者存储如何配置等等），并调用对应的容器平台接口达到这个状态。
2 kubelet启动核心流程
2.1 kubelet的入口函数Main 函数
- 解析参数，对参数的合法性进行判断
- 根据kubeletConfig解析一些特殊的特性所需要配置的参数
- 配置kubeletServer，包括KubeletFlags和KubeletConfiguration两个参数；
- 构造kubeletDeps结构体；
- 启动最终的Run方法：NewKubeletCommand.Run
cmd/kubelet/kubelet.go
func main() {
rand.Seed(time.Now().UTC().UnixNano())
command := app.NewKubeletCommand(server.SetupSignalHandler())
logs.InitLogs()
defer logs.FlushLogs()
if err := command.Execute(); err != nil {
fmt.Fprintf(os.Stderr, "%v\n", err)
os.Exit(1)
}
}
// 初始化命令行
command := app.NewKubeletCommand(server.SetupSignalHandler())
// 执行Execute
err := command.Execute()
// NewKubeletCommand核心逻辑，主要启动NewKubeletCommand.Run
// NewKubeletCommand creates a *cobra.Command object with default parameters
func NewKubeletCommand(stopCh <-chan struct{}) *cobra.Command {
...
cmd := &cobra.Command{
Use: componentKubelet,
Long: `...`,
// The Kubelet has special flag parsing requirements to enforce flag precedence rules,
// so we do all our parsing manually in Run, below.
// DisableFlagParsing=true provides the full set of flags passed to the kubelet in the
// `args` arg to Run, without Cobra's interference.
DisableFlagParsing: true,
Run: func(cmd *cobra.Command, args []string) {
...
// run the kubelet
glog.V(5).Infof("KubeletConfiguration: %#v", kubeletServer.KubeletConfiguration)
//核心跳转逻辑：NewKubeletCommand.Run
if err := Run(kubeletServer, kubeletDeps, stopCh); err != nil {
glog.Fatal(err)
}
},
}
...
return cmd
}
复制代码2.2 Command.Run（<-initForOS）(NewKubeletCommand简称：Command)
- initForOS:initForOS通过对操作系统的判断，如果是windows系统需要做一些预先的特殊处理；run方法即通过传入的kubeDeps参数开始执行启动操作。
- run函数拉开大幕
// Run runs the specified KubeletServer with the given Dependencies. This should never exit.
// The kubeDeps argument may be nil - if so, it is initialized from the settings on KubeletServer.
// Otherwise, the caller is assumed to have set up the Dependencies object and a default one will
// not be generated.
func Run(s *options.KubeletServer, kubeDeps *kubelet.Dependencies, stopCh <-chan struct{}) error {
// To help debugging, immediately log version
klog.Infof("Version: %+v", version.Get())
if err := initForOS(s.KubeletFlags.WindowsService); err != nil {
return fmt.Errorf("failed OS init: %v", err)
}
//进入run方法，开始主要执行对参数的再一次验证，以及新的结构体的初始化。后续开始构建一些
//重要的客户端，包括eventClient主要处理事件的上报，与apiserver打交道；heartbeatClient主
//要处理心跳操作，与之后的PLEG相关；csiClient主要与CSI接口相关。配置完成之后，最终
//进入RunKubelet方法。
if err := run(s, kubeDeps, stopCh); err != nil {
return fmt.Errorf("failed to run Kubelet: %v", err)
}
return nil
}
复制代码2.3 Command.run.run((简称：CRR))
- run(s, kubeDeps, stopCh)：进入run方法，开始构建一些重要的客户端，包括eventClient主要处理事件的上报，与apiserver打交道；heartbeatClient主要处理心跳操作，与之后的PLEG相关；csiClient主要与CSI接口相关。配置完成之后，最终进入RunKubelet方法。
- kubeClient
- dynamicKubeClient
- eventClient
- heartbeatClient
- csiClient
- ContainerManager
- oomAdjuster
- Health check
- 最终执行RunKubelet：通过各种赋值构造了完整的kubeDeps结构体，最后再执行RunKubelet转入后续的kubelet执行流程。
func run(s *options.KubeletServer, kubeDeps *kubelet.Dependencies, stopCh <-chan struct{}) (err error) {
// Set global feature gates based on the value on the initial KubeletServer
err = utilfeature.DefaultMutableFeatureGate.SetFromMap(s.KubeletConfiguration.FeatureGates)
if err != nil {
return err
}
kubeDeps.ContainerManager, err = cm.NewContainerManager(
kubeDeps.Mounter,
kubeDeps.CAdvisorInterface,
cm.NodeConfig{
RuntimeCgroupsName: s.RuntimeCgroups,
SystemCgroupsName: s.SystemCgroups,
KubeletCgroupsName: s.KubeletCgroups,
ContainerRuntime: s.ContainerRuntime,
CgroupsPerQOS: s.CgroupsPerQOS,
CgroupRoot: s.CgroupRoot,
CgroupDriver: s.CgroupDriver,
KubeletRootDir: s.RootDirectory,
ProtectKernelDefaults: s.ProtectKernelDefaults,
NodeAllocatableConfig: cm.NodeAllocatableConfig{
KubeReservedCgroupName: s.KubeReservedCgroup,
SystemReservedCgroupName: s.SystemReservedCgroup,
EnforceNodeAllocatable: sets.NewString(s.EnforceNodeAllocatable...),
KubeReserved: kubeReserved,
SystemReserved: systemReserved,
HardEvictionThresholds: hardEvictionThresholds,
},
QOSReserved: *experimentalQOSReserved,
ExperimentalCPUManagerPolicy: s.CPUManagerPolicy,
ExperimentalCPUManagerReconcilePeriod: s.CPUManagerReconcilePeriod.Duration,
ExperimentalPodPidsLimit: s.PodPidsLimit,
EnforceCPULimits: s.CPUCFSQuota,
CPUCFSQuotaPeriod: s.CPUCFSQuotaPeriod.Duration,
ExperimentalTopologyManagerPolicy: s.TopologyManagerPolicy,
},
s.FailSwapOn,
devicePluginEnabled,
kubeDeps.Recorder)
if err != nil {
return err
}
}
......
// TODO(vmarmol): Do this through container config.
oomAdjuster := kubeDeps.OOMAdjuster
if err := oomAdjuster.ApplyOOMScoreAdj(0, int(s.OOMScoreAdj)); err != nil {
klog.Warning(err)
}
if err := RunKubelet(s, kubeDeps, s.RunOnce); err != nil {
return err
}
复制代码2.4 CRR.RunKubelet(主角登场：简称CRR_RK)
- RunKubelet方法最重要的方法有两个：CreateAndInitKubelet和startKubelet，可以理解为CreateAndInitKubelet为参数的配置，startKubelet为最终的启动。
// RunKubelet is responsible for setting up and running a kubelet. It is used in three different applications:
// 1 Integration tests
// 2 Kubelet binary
// 3 Standalone 'kubernetes' binary
// Eventually, #2 will be replaced with instances of #3
func RunKubelet(kubeServer *options.KubeletServer, kubeDeps *kubelet.Dependencies, runOnce bool) error {
hostname, err := nodeutil.GetHostname(kubeServer.HostnameOverride)
if err != nil {
return err
}
// Query the cloud provider for our node name, default to hostname if kubeDeps.Cloud == nil
nodeName, err := getNodeName(kubeDeps.Cloud, hostname)
if err != nil {
return err
}
// Setup event recorder if required.
makeEventRecorder(kubeDeps, nodeName)
capabilities.Initialize(capabilities.Capabilities{
AllowPrivileged: true,
})
credentialprovider.SetPreferredDockercfgPath(kubeServer.RootDirectory)
klog.V(2).Infof("Using root directory: %v", kubeServer.RootDirectory)
if kubeDeps.OSInterface == nil {
kubeDeps.OSInterface = kubecontainer.RealOS{}
}
//CreateAndInitKubelet方法通过调用NewMainKubelet返回Kubelet结构体
k, err := createAndInitKubelet(&kubeServer.KubeletConfiguration,
kubeDeps,
&kubeServer.ContainerRuntimeOptions,
kubeServer.ContainerRuntime,
kubeServer.RuntimeCgroups,
kubeServer.HostnameOverride,
kubeServer.NodeIP,
kubeServer.ProviderID,
kubeServer.CloudProvider,
kubeServer.CertDirectory,
kubeServer.RootDirectory,
kubeServer.RegisterNode,
kubeServer.RegisterWithTaints,
kubeServer.AllowedUnsafeSysctls,
kubeServer.RemoteRuntimeEndpoint,
kubeServer.RemoteImageEndpoint,
kubeServer.ExperimentalMounterPath,
kubeServer.ExperimentalKernelMemcgNotification,
kubeServer.ExperimentalCheckNodeCapabilitiesBeforeMount,
kubeServer.ExperimentalNodeAllocatableIgnoreEvictionThreshold,
kubeServer.MinimumGCAge,
kubeServer.MaxPerPodContainerCount,
kubeServer.MaxContainerCount,
kubeServer.MasterServiceNamespace,
kubeServer.RegisterSchedulable,
kubeServer.NonMasqueradeCIDR,
kubeServer.KeepTerminatedPodVolumes,
kubeServer.NodeLabels,
kubeServer.SeccompProfileRoot,
kubeServer.BootstrapCheckpointPath,
kubeServer.NodeStatusMaxImages)
if err != nil {
return fmt.Errorf("failed to create kubelet: %v", err)
}
// NewMainKubelet should have set up a pod source config if one didn't exist
// when the builder was run. This is just a precaution.
if kubeDeps.PodConfig == nil {
return fmt.Errorf("failed to create kubelet, pod source config was nil")
}
podCfg := kubeDeps.PodConfig
rlimit.RlimitNumFiles(uint64(kubeServer.MaxOpenFiles))
// process pods and exit.
if runOnce {
if _, err := k.RunOnce(podCfg.Updates()); err != nil {
return fmt.Errorf("runonce failed: %v", err)
}
klog.Info("Started kubelet as runonce")
} else {
//startKubelet方法内部调用了最终的Run方法
startKubelet(k, podCfg, &kubeServer.KubeletConfiguration, kubeDeps, kubeServer.EnableCAdvisorJSONEndpoints, kubeServer.EnableServer)
klog.Info("Started kubelet")
}
return nil
}
复制代码2.5 CRR_RK.CreateAndInitKubelet
- createAndInitKubelet方法中执行的核心函数是NewMainKubelet，NewMainKubelet实例化一个kubelet对象，该部分的具体代码在kubernetes/pkg/kubelet中。
func CreateAndInitKubelet(kubeCfg *kubeletconfiginternal.KubeletConfiguration,
...
nodeStatusMaxImages int32) (k kubelet.Bootstrap, err error) {
// TODO: block until all sources have delivered at least one update to the channel, or break the sync loop
// up into "per source" synchronizations
k, err = kubelet.NewMainKubelet(kubeCfg,
kubeDeps,
crOptions,
containerRuntime,
runtimeCgroups,
hostnameOverride,
nodeIP,
providerID,
cloudProvider,
certDirectory,
rootDirectory,
registerNode,
registerWithTaints,
allowedUnsafeSysctls,
remoteRuntimeEndpoint,
remoteImageEndpoint,
experimentalMounterPath,
experimentalKernelMemcgNotification,
experimentalCheckNodeCapabilitiesBeforeMount,
experimentalNodeAllocatableIgnoreEvictionThreshold,
minimumGCAge,
maxPerPodContainerCount,
maxContainerCount,
masterServiceNamespace,
registerSchedulable,
nonMasqueradeCIDR,
keepTerminatedPodVolumes,
nodeLabels,
seccompProfileRoot,
bootstrapCheckpointPath,
nodeStatusMaxImages)
if err != nil {
return nil, err
}
k.BirthCry()
k.StartGarbageCollection()
return k, nil
}
func NewMainKubelet(kubeCfg *kubeletconfiginternal.KubeletConfiguration,
kubeDeps *Dependencies,
crOptions *config.ContainerRuntimeOptions,
containerRuntime string,
runtimeCgroups string,
hostnameOverride string,
nodeIP string,
providerID string,
cloudProvider string,
certDirectory string,
rootDirectory string,
registerNode bool,
registerWithTaints []api.Taint,
allowedUnsafeSysctls []string,
remoteRuntimeEndpoint string,
remoteImageEndpoint string,
experimentalMounterPath string,
experimentalKernelMemcgNotification bool,
experimentalCheckNodeCapabilitiesBeforeMount bool,
experimentalNodeAllocatableIgnoreEvictionThreshold bool,
minimumGCAge metav1.Duration,
maxPerPodContainerCount int32,
maxContainerCount int32,
masterServiceNamespace string,
registerSchedulable bool,
nonMasqueradeCIDR string,
keepTerminatedPodVolumes bool,
nodeLabels map[string]string,
seccompProfileRoot string,
bootstrapCheckpointPath string,
nodeStatusMaxImages int32) (*Kubelet, error) {
if rootDirectory == "" {
return nil, fmt.Errorf("invalid root directory %q", rootDirectory)
}
if kubeCfg.SyncFrequency.Duration <= 0 {
return nil, fmt.Errorf("invalid sync frequency %d", kubeCfg.SyncFrequency.Duration)
}
if kubeCfg.MakeIPTablesUtilChains {
if kubeCfg.IPTablesMasqueradeBit > 31 || kubeCfg.IPTablesMasqueradeBit < 0 {
return nil, fmt.Errorf("iptables-masquerade-bit is not valid. Must be within [0, 31]")
}
if kubeCfg.IPTablesDropBit > 31 || kubeCfg.IPTablesDropBit < 0 {
return nil, fmt.Errorf("iptables-drop-bit is not valid. Must be within [0, 31]")
}
if kubeCfg.IPTablesDropBit == kubeCfg.IPTablesMasqueradeBit {
return nil, fmt.Errorf("iptables-masquerade-bit and iptables-drop-bit must be different")
}
}
hostname, err := nodeutil.GetHostname(hostnameOverride)
if err != nil {
return nil, err
}
// Query the cloud provider for our node name, default to hostname
nodeName := types.NodeName(hostname)
if kubeDeps.Cloud != nil {
var err error
instances, ok := kubeDeps.Cloud.Instances()
if !ok {
return nil, fmt.Errorf("failed to get instances from cloud provider")
}
nodeName, err = instances.CurrentNodeName(context.TODO(), hostname)
if err != nil {
return nil, fmt.Errorf("error fetching current instance name from cloud provider: %v", err)
}
klog.V(2).Infof("cloud provider determined current node name to be %s", nodeName)
}
if kubeDeps.PodConfig == nil {
var err error
kubeDeps.PodConfig, err = makePodSourceConfig(kubeCfg, kubeDeps, nodeName, bootstrapCheckpointPath)
if err != nil {
return nil, err
}
}
containerGCPolicy := kubecontainer.ContainerGCPolicy{
MinAge: minimumGCAge.Duration,
MaxPerPodContainer: int(maxPerPodContainerCount),
MaxContainers: int(maxContainerCount),
}
daemonEndpoints := &v1.NodeDaemonEndpoints{
KubeletEndpoint: v1.DaemonEndpoint{Port: kubeCfg.Port},
}
imageGCPolicy := images.ImageGCPolicy{
MinAge: kubeCfg.ImageMinimumGCAge.Duration,
HighThresholdPercent: int(kubeCfg.ImageGCHighThresholdPercent),
LowThresholdPercent: int(kubeCfg.ImageGCLowThresholdPercent),
}
enforceNodeAllocatable := kubeCfg.EnforceNodeAllocatable
if experimentalNodeAllocatableIgnoreEvictionThreshold {
// Do not provide kubeCfg.EnforceNodeAllocatable to eviction threshold parsing if we are not enforcing Evictions
enforceNodeAllocatable = []string{}
}
thresholds, err := eviction.ParseThresholdConfig(enforceNodeAllocatable, kubeCfg.EvictionHard, kubeCfg.EvictionSoft, kubeCfg.EvictionSoftGracePeriod, kubeCfg.EvictionMinimumReclaim)
if err != nil {
return nil, err
}
evictionConfig := eviction.Config{
PressureTransitionPeriod: kubeCfg.EvictionPressureTransitionPeriod.Duration,
MaxPodGracePeriodSeconds: int64(kubeCfg.EvictionMaxPodGracePeriod),
Thresholds: thresholds,
KernelMemcgNotification: experimentalKernelMemcgNotification,
PodCgroupRoot: kubeDeps.ContainerManager.GetPodCgroupRoot(),
}
serviceIndexer := cache.NewIndexer(cache.MetaNamespaceKeyFunc, cache.Indexers{cache.NamespaceIndex: cache.MetaNamespaceIndexFunc})
if kubeDeps.KubeClient != nil {
serviceLW := cache.NewListWatchFromClient(kubeDeps.KubeClient.CoreV1().RESTClient(), "services", metav1.NamespaceAll, fields.Everything())
r := cache.NewReflector(serviceLW, &v1.Service{}, serviceIndexer, 0)
go r.Run(wait.NeverStop)
}
serviceLister := corelisters.NewServiceLister(serviceIndexer)
nodeIndexer := cache.NewIndexer(cache.MetaNamespaceKeyFunc, cache.Indexers{})
if kubeDeps.KubeClient != nil {
fieldSelector := fields.Set{api.ObjectNameField: string(nodeName)}.AsSelector()
nodeLW := cache.NewListWatchFromClient(kubeDeps.KubeClient.CoreV1().RESTClient(), "nodes", metav1.NamespaceAll, fieldSelector)
r := cache.NewReflector(nodeLW, &v1.Node{}, nodeIndexer, 0)
go r.Run(wait.NeverStop)
}
nodeInfo := &CachedNodeInfo{NodeLister: corelisters.NewNodeLister(nodeIndexer)}
// TODO: get the real node object of ourself,
// and use the real node name and UID.
// TODO: what is namespace for node?
nodeRef := &v1.ObjectReference{
Kind: "Node",
Name: string(nodeName),
UID: types.UID(nodeName),
Namespace: "",
}
containerRefManager := kubecontainer.NewRefManager()
oomWatcher := oomwatcher.NewWatcher(kubeDeps.Recorder)
clusterDNS := make([]net.IP, 0, len(kubeCfg.ClusterDNS))
for _, ipEntry := range kubeCfg.ClusterDNS {
ip := net.ParseIP(ipEntry)
if ip == nil {
klog.Warningf("Invalid clusterDNS ip '%q'", ipEntry)
} else {
clusterDNS = append(clusterDNS, ip)
}
}
httpClient := &http.Client{}
parsedNodeIP := net.ParseIP(nodeIP)
protocol := utilipt.ProtocolIpv4
if parsedNodeIP != nil && parsedNodeIP.To4() == nil {
klog.V(0).Infof("IPv6 node IP (%s), assume IPv6 operation", nodeIP)
protocol = utilipt.ProtocolIpv6
}
klet := &Kubelet{
hostname: hostname,
hostnameOverridden: len(hostnameOverride) > 0,
nodeName: nodeName,
kubeClient: kubeDeps.KubeClient,
heartbeatClient: kubeDeps.HeartbeatClient,
onRepeatedHeartbeatFailure: kubeDeps.OnHeartbeatFailure,
rootDirectory: rootDirectory,
resyncInterval: kubeCfg.SyncFrequency.Duration,
sourcesReady: config.NewSourcesReady(kubeDeps.PodConfig.SeenAllSources),
registerNode: registerNode,
registerWithTaints: registerWithTaints,
registerSchedulable: registerSchedulable,
dnsConfigurer: dns.NewConfigurer(kubeDeps.Recorder, nodeRef, parsedNodeIP, clusterDNS, kubeCfg.ClusterDomain, kubeCfg.ResolverConfig),
serviceLister: serviceLister,
nodeInfo: nodeInfo,
masterServiceNamespace: masterServiceNamespace,
streamingConnectionIdleTimeout: kubeCfg.StreamingConnectionIdleTimeout.Duration,
recorder: kubeDeps.Recorder,
cadvisor: kubeDeps.CAdvisorInterface,
cloud: kubeDeps.Cloud,
externalCloudProvider: cloudprovider.IsExternal(cloudProvider),
providerID: providerID,
nodeRef: nodeRef,
nodeLabels: nodeLabels,
nodeStatusUpdateFrequency: kubeCfg.NodeStatusUpdateFrequency.Duration,
nodeStatusReportFrequency: kubeCfg.NodeStatusReportFrequency.Duration,
os: kubeDeps.OSInterface,
oomWatcher: oomWatcher,
cgroupsPerQOS: kubeCfg.CgroupsPerQOS,
cgroupRoot: kubeCfg.CgroupRoot,
mounter: kubeDeps.Mounter,
hostutil: kubeDeps.HostUtil,
subpather: kubeDeps.Subpather,
maxPods: int(kubeCfg.MaxPods),
podsPerCore: int(kubeCfg.PodsPerCore),
syncLoopMonitor: atomic.Value{},
daemonEndpoints: daemonEndpoints,
containerManager: kubeDeps.ContainerManager,
containerRuntimeName: containerRuntime,
redirectContainerStreaming: crOptions.RedirectContainerStreaming,
nodeIP: parsedNodeIP,
nodeIPValidator: validateNodeIP,
clock: clock.RealClock{},
enableControllerAttachDetach: kubeCfg.EnableControllerAttachDetach,
iptClient: utilipt.New(utilexec.New(), utildbus.New(), protocol),
makeIPTablesUtilChains: kubeCfg.MakeIPTablesUtilChains,
iptablesMasqueradeBit: int(kubeCfg.IPTablesMasqueradeBit),
iptablesDropBit: int(kubeCfg.IPTablesDropBit),
experimentalHostUserNamespaceDefaulting: utilfeature.DefaultFeatureGate.Enabled(features.ExperimentalHostUserNamespaceDefaultingGate),
keepTerminatedPodVolumes: keepTerminatedPodVolumes,
nodeStatusMaxImages: nodeStatusMaxImages,
}
if klet.cloud != nil {
klet.cloudResourceSyncManager = cloudresource.NewSyncManager(klet.cloud, nodeName, klet.nodeStatusUpdateFrequency)
}
var secretManager secret.Manager
var configMapManager configmap.Manager
switch kubeCfg.ConfigMapAndSecretChangeDetectionStrategy {
case kubeletconfiginternal.WatchChangeDetectionStrategy:
secretManager = secret.NewWatchingSecretManager(kubeDeps.KubeClient)
configMapManager = configmap.NewWatchingConfigMapManager(kubeDeps.KubeClient)
case kubeletconfiginternal.TTLCacheChangeDetectionStrategy:
secretManager = secret.NewCachingSecretManager(
kubeDeps.KubeClient, manager.GetObjectTTLFromNodeFunc(klet.GetNode))
configMapManager = configmap.NewCachingConfigMapManager(
kubeDeps.KubeClient, manager.GetObjectTTLFromNodeFunc(klet.GetNode))
case kubeletconfiginternal.GetChangeDetectionStrategy:
secretManager = secret.NewSimpleSecretManager(kubeDeps.KubeClient)
configMapManager = configmap.NewSimpleConfigMapManager(kubeDeps.KubeClient)
default:
return nil, fmt.Errorf("unknown configmap and secret manager mode: %v", kubeCfg.ConfigMapAndSecretChangeDetectionStrategy)
}
klet.secretManager = secretManager
klet.configMapManager = configMapManager
if klet.experimentalHostUserNamespaceDefaulting {
klog.Infof("Experimental host user namespace defaulting is enabled.")
}
machineInfo, err := klet.cadvisor.MachineInfo()
if err != nil {
return nil, err
}
klet.machineInfo = machineInfo
imageBackOff := flowcontrol.NewBackOff(backOffPeriod, MaxContainerBackOff)
klet.livenessManager = proberesults.NewManager()
klet.podCache = kubecontainer.NewCache()
var checkpointManager checkpointmanager.CheckpointManager
if bootstrapCheckpointPath != "" {
checkpointManager, err = checkpointmanager.NewCheckpointManager(bootstrapCheckpointPath)
if err != nil {
return nil, fmt.Errorf("failed to initialize checkpoint manager: %+v", err)
}
}
// podManager is also responsible for keeping secretManager and configMapManager contents up-to-date.
klet.podManager = kubepod.NewBasicPodManager(kubepod.NewBasicMirrorClient(klet.kubeClient), secretManager, configMapManager, checkpointManager)
klet.statusManager = status.NewManager(klet.kubeClient, klet.podManager, klet)
if remoteRuntimeEndpoint != "" {
// remoteImageEndpoint is same as remoteRuntimeEndpoint if not explicitly specified
if remoteImageEndpoint == "" {
remoteImageEndpoint = remoteRuntimeEndpoint
}
}
// TODO: These need to become arguments to a standalone docker shim.
pluginSettings := dockershim.NetworkPluginSettings{
HairpinMode: kubeletconfiginternal.HairpinMode(kubeCfg.HairpinMode),
NonMasqueradeCIDR: nonMasqueradeCIDR,
PluginName: crOptions.NetworkPluginName,
PluginConfDir: crOptions.CNIConfDir,
PluginBinDirString: crOptions.CNIBinDir,
PluginCacheDir: crOptions.CNICacheDir,
MTU: int(crOptions.NetworkPluginMTU),
}
klet.resourceAnalyzer = serverstats.NewResourceAnalyzer(klet, kubeCfg.VolumeStatsAggPeriod.Duration)
// if left at nil, that means it is unneeded
var legacyLogProvider kuberuntime.LegacyLogProvider
switch containerRuntime {
case kubetypes.DockerContainerRuntime:
// Create and start the CRI shim running as a grpc server.
streamingConfig := getStreamingConfig(kubeCfg, kubeDeps, crOptions)
ds, err := dockershim.NewDockerService(kubeDeps.DockerClientConfig, crOptions.PodSandboxImage, streamingConfig,
&pluginSettings, runtimeCgroups, kubeCfg.CgroupDriver, crOptions.DockershimRootDirectory, !crOptions.RedirectContainerStreaming)
if err != nil {
return nil, err
}
if crOptions.RedirectContainerStreaming {
klet.criHandler = ds
}
// The unix socket for kubelet <-> dockershim communication.
klog.V(5).Infof("RemoteRuntimeEndpoint: %q, RemoteImageEndpoint: %q",
remoteRuntimeEndpoint,
remoteImageEndpoint)
klog.V(2).Infof("Starting the GRPC server for the docker CRI shim.")
server := dockerremote.NewDockerServer(remoteRuntimeEndpoint, ds)
if err := server.Start(); err != nil {
return nil, err
}
// Create dockerLegacyService when the logging driver is not supported.
supported, err := ds.IsCRISupportedLogDriver()
if err != nil {
return nil, err
}
if !supported {
klet.dockerLegacyService = ds
legacyLogProvider = ds
}
case kubetypes.RemoteContainerRuntime:
// No-op.
break
default:
return nil, fmt.Errorf("unsupported CRI runtime: %q", containerRuntime)
}
runtimeService, imageService, err := getRuntimeAndImageServices(remoteRuntimeEndpoint, remoteImageEndpoint, kubeCfg.RuntimeRequestTimeout)
if err != nil {
return nil, err
}
klet.runtimeService = runtimeService
if utilfeature.DefaultFeatureGate.Enabled(features.RuntimeClass) && kubeDeps.KubeClient != nil {
klet.runtimeClassManager = runtimeclass.NewManager(kubeDeps.KubeClient)
}
runtime, err := kuberuntime.NewKubeGenericRuntimeManager(
kubecontainer.FilterEventRecorder(kubeDeps.Recorder),
klet.livenessManager,
seccompProfileRoot,
containerRefManager,
machineInfo,
klet,
kubeDeps.OSInterface,
klet,
httpClient,
imageBackOff,
kubeCfg.SerializeImagePulls,
float32(kubeCfg.RegistryPullQPS),
int(kubeCfg.RegistryBurst),
kubeCfg.CPUCFSQuota,
kubeCfg.CPUCFSQuotaPeriod,
runtimeService,
imageService,
kubeDeps.ContainerManager.InternalContainerLifecycle(),
legacyLogProvider,
klet.runtimeClassManager,
)
if err != nil {
return nil, err
}
klet.containerRuntime = runtime
klet.streamingRuntime = runtime
klet.runner = runtime
runtimeCache, err := kubecontainer.NewRuntimeCache(klet.containerRuntime)
if err != nil {
return nil, err
}
klet.runtimeCache = runtimeCache
if cadvisor.UsingLegacyCadvisorStats(containerRuntime, remoteRuntimeEndpoint) {
klet.StatsProvider = stats.NewCadvisorStatsProvider(
klet.cadvisor,
klet.resourceAnalyzer,
klet.podManager,
klet.runtimeCache,
klet.containerRuntime,
klet.statusManager)
} else {
klet.StatsProvider = stats.NewCRIStatsProvider(
klet.cadvisor,
klet.resourceAnalyzer,
klet.podManager,
klet.runtimeCache,
runtimeService,
imageService,
stats.NewLogMetricsService(),
kubecontainer.RealOS{})
}
klet.pleg = pleg.NewGenericPLEG(klet.containerRuntime, plegChannelCapacity, plegRelistPeriod, klet.podCache, clock.RealClock{})
klet.runtimeState = newRuntimeState(maxWaitForContainerRuntime)
klet.runtimeState.addHealthCheck("PLEG", klet.pleg.Healthy)
if _, err := klet.updatePodCIDR(kubeCfg.PodCIDR); err != nil {
klog.Errorf("Pod CIDR update failed %v", err)
}
// setup containerGC
containerGC, err := kubecontainer.NewContainerGC(klet.containerRuntime, containerGCPolicy, klet.sourcesReady)
if err != nil {
return nil, err
}
klet.containerGC = containerGC
klet.containerDeletor = newPodContainerDeletor(klet.containerRuntime, integer.IntMax(containerGCPolicy.MaxPerPodContainer, minDeadContainerInPod))
// setup imageManager
imageManager, err := images.NewImageGCManager(klet.containerRuntime, klet.StatsProvider, kubeDeps.Recorder, nodeRef, imageGCPolicy, crOptions.PodSandboxImage)
if err != nil {
return nil, fmt.Errorf("failed to initialize image manager: %v", err)
}
klet.imageManager = imageManager
if containerRuntime == kubetypes.RemoteContainerRuntime && utilfeature.DefaultFeatureGate.Enabled(features.CRIContainerLogRotation) {
// setup containerLogManager for CRI container runtime
containerLogManager, err := logs.NewContainerLogManager(
klet.runtimeService,
kubeCfg.ContainerLogMaxSize,
int(kubeCfg.ContainerLogMaxFiles),
)
if err != nil {
return nil, fmt.Errorf("failed to initialize container log manager: %v", err)
}
klet.containerLogManager = containerLogManager
} else {
klet.containerLogManager = logs.NewStubContainerLogManager()
}
if kubeCfg.ServerTLSBootstrap && kubeDeps.TLSOptions != nil && utilfeature.DefaultFeatureGate.Enabled(features.RotateKubeletServerCertificate) {
klet.serverCertificateManager, err = kubeletcertificate.NewKubeletServerCertificateManager(klet.kubeClient, kubeCfg, klet.nodeName, klet.getLastObservedNodeAddresses, certDirectory)
if err != nil {
return nil, fmt.Errorf("failed to initialize certificate manager: %v", err)
}
kubeDeps.TLSOptions.Config.GetCertificate = func(*tls.ClientHelloInfo) (*tls.Certificate, error) {
cert := klet.serverCertificateManager.Current()
if cert == nil {
return nil, fmt.Errorf("no serving certificate available for the kubelet")
}
return cert, nil
}
}
klet.probeManager = prober.NewManager(
klet.statusManager,
klet.livenessManager,
klet.runner,
containerRefManager,
kubeDeps.Recorder)
tokenManager := token.NewManager(kubeDeps.KubeClient)
// NewInitializedVolumePluginMgr initializes some storageErrors on the Kubelet runtimeState (in csi_plugin.go init)
// which affects node ready status. This function must be called before Kubelet is initialized so that the Node
// ReadyState is accurate with the storage state.
klet.volumePluginMgr, err =
NewInitializedVolumePluginMgr(klet, secretManager, configMapManager, tokenManager, kubeDeps.VolumePlugins, kubeDeps.DynamicPluginProber)
if err != nil {
return nil, err
}
klet.pluginManager = pluginmanager.NewPluginManager(
klet.getPluginsRegistrationDir(), /* sockDir */
klet.getPluginsDir(), /* deprecatedSockDir */
kubeDeps.Recorder,
)
// If the experimentalMounterPathFlag is set, we do not want to
// check node capabilities since the mount path is not the default
if len(experimentalMounterPath) != 0 {
experimentalCheckNodeCapabilitiesBeforeMount = false
// Replace the nameserver in containerized-mounter's rootfs/etc/resolve.conf with kubelet.ClusterDNS
// so that service name could be resolved
klet.dnsConfigurer.SetupDNSinContainerizedMounter(experimentalMounterPath)
}
// setup volumeManager
klet.volumeManager = volumemanager.NewVolumeManager(
kubeCfg.EnableControllerAttachDetach,
nodeName,
klet.podManager,
klet.statusManager,
klet.kubeClient,
klet.volumePluginMgr,
klet.containerRuntime,
kubeDeps.Mounter,
kubeDeps.HostUtil,
klet.getPodsDir(),
kubeDeps.Recorder,
experimentalCheckNodeCapabilitiesBeforeMount,
keepTerminatedPodVolumes,
volumepathhandler.NewBlockVolumePathHandler())
klet.reasonCache = NewReasonCache()
klet.workQueue = queue.NewBasicWorkQueue(klet.clock)
klet.podWorkers = newPodWorkers(klet.syncPod, kubeDeps.Recorder, klet.workQueue, klet.resyncInterval, backOffPeriod, klet.podCache)
klet.backOff = flowcontrol.NewBackOff(backOffPeriod, MaxContainerBackOff)
klet.podKillingCh = make(chan *kubecontainer.PodPair, podKillingChannelCapacity)
// setup eviction manager
evictionManager, evictionAdmitHandler := eviction.NewManager(klet.resourceAnalyzer, evictionConfig, killPodNow(klet.podWorkers, kubeDeps.Recorder), klet.podManager.GetMirrorPodByPod, klet.imageManager, klet.containerGC, kubeDeps.Recorder, nodeRef, klet.clock)
klet.evictionManager = evictionManager
klet.admitHandlers.AddPodAdmitHandler(evictionAdmitHandler)
if utilfeature.DefaultFeatureGate.Enabled(features.Sysctls) {
// add sysctl admission
runtimeSupport, err := sysctl.NewRuntimeAdmitHandler(klet.containerRuntime)
if err != nil {
return nil, err
}
// Safe, whitelisted sysctls can always be used as unsafe sysctls in the spec.
// Hence, we concatenate those two lists.
safeAndUnsafeSysctls := append(sysctlwhitelist.SafeSysctlWhitelist(), allowedUnsafeSysctls...)
sysctlsWhitelist, err := sysctl.NewWhitelist(safeAndUnsafeSysctls)
if err != nil {
return nil, err
}
klet.admitHandlers.AddPodAdmitHandler(runtimeSupport)
klet.admitHandlers.AddPodAdmitHandler(sysctlsWhitelist)
}
// enable active deadline handler
activeDeadlineHandler, err := newActiveDeadlineHandler(klet.statusManager, kubeDeps.Recorder, klet.clock)
if err != nil {
return nil, err
}
klet.AddPodSyncLoopHandler(activeDeadlineHandler)
klet.AddPodSyncHandler(activeDeadlineHandler)
if utilfeature.DefaultFeatureGate.Enabled(features.TopologyManager) {
klet.admitHandlers.AddPodAdmitHandler(klet.containerManager.GetTopologyPodAdmitHandler())
}
criticalPodAdmissionHandler := preemption.NewCriticalPodAdmissionHandler(klet.GetActivePods, killPodNow(klet.podWorkers, kubeDeps.Recorder), kubeDeps.Recorder)
klet.admitHandlers.AddPodAdmitHandler(lifecycle.NewPredicateAdmitHandler(klet.getNodeAnyWay, criticalPodAdmissionHandler, klet.containerManager.UpdatePluginResources))
// apply functional Option's
for _, opt := range kubeDeps.Options {
opt(klet)
}
klet.appArmorValidator = apparmor.NewValidator(containerRuntime)
klet.softAdmitHandlers.AddPodAdmitHandler(lifecycle.NewAppArmorAdmitHandler(klet.appArmorValidator))
klet.softAdmitHandlers.AddPodAdmitHandler(lifecycle.NewNoNewPrivsAdmitHandler(klet.containerRuntime))
if utilfeature.DefaultFeatureGate.Enabled(features.NodeLease) {
klet.nodeLeaseController = nodelease.NewController(klet.clock, klet.heartbeatClient, string(klet.nodeName), kubeCfg.NodeLeaseDurationSeconds, klet.onRepeatedHeartbeatFailure)
}
klet.softAdmitHandlers.AddPodAdmitHandler(lifecycle.NewProcMountAdmitHandler(klet.containerRuntime))
// Finally, put the most recent version of the config on the Kubelet, so
// people can see how it was configured.
klet.kubeletConfiguration = *kubeCfg
// Generating the status funcs should be the last thing we do,
// since this relies on the rest of the Kubelet having been constructed.
klet.setNodeStatusFuncs = klet.defaultNodeStatusFuncs()
return klet, nil
}
复制代码2.6 CRR_RK.startKubelet(至此方休)
- 最终进入核心函数：Kubelet.run
- 1、volumeManager，volume相关管理；
- 2、syncNodeStatus，定时同步Node状态；
- 3、updateRuntimeUp，定时更新Runtime状态；
- 4、syncNetworkUtil，定时同步网络状态；
- 5、podKiller，定时清理死亡的pod；
- 6、statusManager，pod状态管理；
- 7、probeManager，pod探针管理；
- 8、启动PLEG；
- 9、syncLoop，最重要的主进程，不停监听外部数据的变化执行pod的相应操作。 至此，kubelet启动过程完成。启动主要完成的任务就是参数的配置和多个任务的启动，通过构造一个循环进程不停监听外部事件的变化，执行对应的pod处理工作，这也就是kubelet所需要负责的任务。
func startKubelet(k kubelet.Bootstrap, podCfg *config.PodConfig, kubeCfg *kubeletconfiginternal.KubeletConfiguration, kubeDeps *kubelet.Dependencies, enableCAdvisorJSONEndpoints, enableServer bool) {
// start the kubelet
go wait.Until(func() {
k.Run(podCfg.Updates())
}, 0, wait.NeverStop)
// start the kubelet server
if enableServer {
go k.ListenAndServe(net.ParseIP(kubeCfg.Address), uint(kubeCfg.Port), kubeDeps.TLSOptions, kubeDeps.Auth, enableCAdvisorJSONEndpoints, kubeCfg.EnableDebuggingHandlers, kubeCfg.EnableContentionProfiling)
}
if kubeCfg.ReadOnlyPort > 0 {
go k.ListenAndServeReadOnly(net.ParseIP(kubeCfg.Address), uint(kubeCfg.ReadOnlyPort), enableCAdvisorJSONEndpoints)
}
if utilfeature.DefaultFeatureGate.Enabled(features.KubeletPodResources) {
go k.ListenAndServePodResources()
}
}
// Run starts the kubelet reacting to config updates
func (kl *Kubelet) Run(updates <-chan kubetypes.PodUpdate) {
if kl.logServer == nil {
kl.logServer = http.StripPrefix("/logs/", http.FileServer(http.Dir("/var/log/")))
}
if kl.kubeClient == nil {
klog.Warning("No api server defined - no node status update will be sent.")
}
// Start the cloud provider sync manager
if kl.cloudResourceSyncManager != nil {
go kl.cloudResourceSyncManager.Run(wait.NeverStop)
}
if err := kl.initializeModules(); err != nil {
kl.recorder.Eventf(kl.nodeRef, v1.EventTypeWarning, events.KubeletSetupFailed, err.Error())
klog.Fatal(err)
}
// Start volume manager
go kl.volumeManager.Run(kl.sourcesReady, wait.NeverStop)
if kl.kubeClient != nil {
// Start syncing node status immediately, this may set up things the runtime needs to run.
go wait.Until(kl.syncNodeStatus, kl.nodeStatusUpdateFrequency, wait.NeverStop)
go kl.fastStatusUpdateOnce()
// start syncing lease
if utilfeature.DefaultFeatureGate.Enabled(features.NodeLease) {
go kl.nodeLeaseController.Run(wait.NeverStop)
}
}
go wait.Until(kl.updateRuntimeUp, 5*time.Second, wait.NeverStop)
// Start loop to sync iptables util rules
if kl.makeIPTablesUtilChains {
go wait.Until(kl.syncNetworkUtil, 1*time.Minute, wait.NeverStop)
}
// Start a goroutine responsible for killing pods (that are not properly
// handled by pod workers).
go wait.Until(kl.podKiller, 1*time.Second, wait.NeverStop)
// Start component sync loops.
kl.statusManager.Start()
kl.probeManager.Start()
// Start syncing RuntimeClasses if enabled.
if kl.runtimeClassManager != nil {
kl.runtimeClassManager.Start(wait.NeverStop)
}
// Start the pod lifecycle event generator.
kl.pleg.Start()
kl.syncLoop(updates, kl)
}
复制代码- 通过长驻进程的方式运行k.Run，
3 总结
- cmd部分调用流程如下：Main-->NewKubeletCommand-->Run(kubeletServer, kubeletDeps, stopCh)-->run(s *options.KubeletServer, kubeDeps ..., stopCh ...)--> RunKubelet(s, kubeDeps, s.RunOnce)-->startKubelet-->k.Run(podCfg.Updates())-->pkg/kubelet。
- RunKubelet(s, kubeDeps, s.RunOnce)-->CreateAndInitKubelet-->kubelet.NewMainKubelet-->pkg/kubelet。
专注于大数据及容器云核心技术解密，可提供全栈的大数据+云原生平台咨询方案，请持续关注本套博客。如有任何学术交流，可随时联系。更多内容请关注《数据云技术社区》公众号。