一文讀懂 Kubernetes APIServer 原理

時間 2021-02-14

標籤 node git github 數據庫 bootstrap 後端 api 數組 restful app 欄目 Git 简体版

原文原文鏈接

前言

整個Kubernetes技術體系由聲明式API以及Controller構成，而kube-apiserver是Kubernetes的聲明式api server，併爲其它組件交互提供了橋樑。所以加深對kube-apiserver的理解就顯得相當重要了。node

總體組件功能

kube-apiserver做爲整個Kubernetes集羣操做etcd的惟一入口，負責Kubernetes各資源的認證&鑑權，校驗以及CRUD等操做，提供RESTful APIs，供其它組件調用：git

kube-apiserver包含三種APIServer：github

aggregatorServer：負責處理 apiregistration.k8s.io 組下的APIService資源請求，同時未來自用戶的請求攔截轉發給aggregated server(AA)
kubeAPIServer：負責對請求的一些通用處理，包括：認證、鑑權以及各個內建資源(pod, deployment，service and etc)的REST服務等
apiExtensionsServer：負責CustomResourceDefinition（CRD）apiResources以及apiVersions的註冊，同時處理CRD以及相應CustomResource（CR）的REST請求(若是對應CR不能被處理的話則會返回404)，也是apiserver Delegation的最後一環

另外還包括bootstrap-controller，主要負責Kubernetes default apiserver service的建立以及管理。數據庫

接下來將對上述組件進行概覽性總結。bootstrap

bootstrap-controller

apiserver bootstrap-controller建立&運行邏輯在k8s.io/kubernetes/pkg/master目錄
bootstrap-controller主要用於建立以及維護內部kubernetes default apiserver service
kubernetes default apiserver service spec.selector爲空，這是default apiserver service與其它正常service的最大區別，代表了這個特殊的service對應的endpoints不禁endpoints controller控制，而是直接受kube-apiserver bootstrap-controller管理(maintained by this code, not by the pod selector)
bootstrap-controller的幾個主要功能以下：後端
- 建立 default、kube-system 和 kube-public 以及 kube-node-lease 命名空間
- 建立&維護kubernetes default apiserver service以及對應的endpoint
- 提供基於Service ClusterIP的檢查及修復功能(--service-cluster-ip-range指定範圍)
- 提供基於Service NodePort的檢查及修復功能(--service-node-port-range指定範圍)

// k8s.io/kubernetes/pkg/master/controller.go:142
// Start begins the core controller loops that must exist for bootstrapping
// a cluster.
func (c *Controller) Start() {
    if c.runner != nil {
        return
    }
    // Reconcile during first run removing itself until server is ready.
    endpointPorts := createEndpointPortSpec(c.PublicServicePort, "https", c.ExtraEndpointPorts)
    if err := c.EndpointReconciler.RemoveEndpoints(kubernetesServiceName, c.PublicIP, endpointPorts); err != nil {
        klog.Errorf("Unable to remove old endpoints from kubernetes service: %v", err)
    }
    repairClusterIPs := servicecontroller.NewRepair(c.ServiceClusterIPInterval, c.ServiceClient, c.EventClient, &c.ServiceClusterIPRange, c.ServiceClusterIPRegistry, &c.SecondaryServiceClusterIPRange, c.SecondaryServiceClusterIPRegistry)
    repairNodePorts := portallocatorcontroller.NewRepair(c.ServiceNodePortInterval, c.ServiceClient, c.EventClient, c.ServiceNodePortRange, c.ServiceNodePortRegistry)
    // run all of the controllers once prior to returning from Start.
    if err := repairClusterIPs.RunOnce(); err != nil {
        // If we fail to repair cluster IPs apiserver is useless. We should restart and retry.
        klog.Fatalf("Unable to perform initial IP allocation check: %v", err)
    }
    if err := repairNodePorts.RunOnce(); err != nil {
        // If we fail to repair node ports apiserver is useless. We should restart and retry.
        klog.Fatalf("Unable to perform initial service nodePort check: %v", err)
    }
    // 按期執行bootstrap controller主要的四個功能(reconciliation)  
    c.runner = async.NewRunner(c.RunKubernetesNamespaces, c.RunKubernetesService, repairClusterIPs.RunUntil, repairNodePorts.RunUntil)
    c.runner.Start()
}

更多代碼原理詳情，參考 kubernetes-reading-notes 。api

kubeAPIServer

KubeAPIServer主要提供對內建API Resources的操做請求，爲Kubernetes中各API Resources註冊路由信息，同時暴露RESTful API，使集羣中以及集羣外的服務均可以經過RESTful API操做Kubernetes中的資源數組

另外，kubeAPIServer是整個Kubernetes apiserver的核心，下面將要講述的aggregatorServer以及apiExtensionsServer都是創建在kubeAPIServer基礎上進行擴展的(補充了Kubernetes對用戶自定義資源的能力支持)restful

kubeAPIServer最核心的功能是爲Kubernetes內置資源添加路由，以下：app

調用 m.InstallLegacyAPI 將核心 API Resources添加到路由中，在apiserver中便是以 /api 開頭的 resource；
調用 m.InstallAPIs 將擴展的 API Resources添加到路由中，在apiserver中便是以 /apis 開頭的 resource；

// k8s.io/kubernetes/pkg/master/master.go:332
// New returns a new instance of Master from the given config.
// Certain config fields will be set to a default value if unset.
// Certain config fields must be specified, including:
//   KubeletClientConfig
func (c completedConfig) New(delegationTarget genericapiserver.DelegationTarget) (*Master, error) {
    ...
    // 安裝 LegacyAPI(core API)
    // install legacy rest storage
    if c.ExtraConfig.APIResourceConfigSource.VersionEnabled(apiv1.SchemeGroupVersion) {
        legacyRESTStorageProvider := corerest.LegacyRESTStorageProvider{
            StorageFactory:              c.ExtraConfig.StorageFactory,
            ProxyTransport:              c.ExtraConfig.ProxyTransport,
            KubeletClientConfig:         c.ExtraConfig.KubeletClientConfig,
            EventTTL:                    c.ExtraConfig.EventTTL,
            ServiceIPRange:              c.ExtraConfig.ServiceIPRange,
            SecondaryServiceIPRange:     c.ExtraConfig.SecondaryServiceIPRange,
            ServiceNodePortRange:        c.ExtraConfig.ServiceNodePortRange,
            LoopbackClientConfig:        c.GenericConfig.LoopbackClientConfig,
            ServiceAccountIssuer:        c.ExtraConfig.ServiceAccountIssuer,
            ServiceAccountMaxExpiration: c.ExtraConfig.ServiceAccountMaxExpiration,
            APIAudiences:                c.GenericConfig.Authentication.APIAudiences,
        }
        if err := m.InstallLegacyAPI(&c, c.GenericConfig.RESTOptionsGetter, legacyRESTStorageProvider); err != nil {
            return nil, err
        }
    }
    ...
    // 安裝 APIs(named groups apis)
    if err := m.InstallAPIs(c.ExtraConfig.APIResourceConfigSource, c.GenericConfig.RESTOptionsGetter, restStorageProviders...); err != nil {
        return nil, err
    }
    ...
    return m, nil
}

整個kubeAPIServer提供了三類API Resource接口：

core group：主要在 /api/v1 下；
named groups：其 path 爲 /apis/$GROUP/$VERSION；
系統狀態的一些 API：如/metrics 、/version 等；

而API的URL大體以 /apis/{group}/{version}/namespaces/{namespace}/resource/{name} 組成，結構以下圖所示：

kubeAPIServer會爲每種API資源建立對應的RESTStorage，RESTStorage的目的是將每種資源的訪問路徑及其後端存儲的操做對應起來：經過構造的REST Storage實現的接口判斷該資源能夠執行哪些操做（如：create、update等），將其對應的操做存入到action中，每個操做對應一個標準的REST method，如create對應REST method爲POST，而update對應REST method爲PUT。最終根據actions數組依次遍歷，對每個操做添加一個handler(handler對應REST Storage實現的相關接口)，並註冊到route，最終對外提供RESTful API，以下：

// m.GenericAPIServer.InstallLegacyAPIGroup --> s.installAPIResources --> apiGroupVersion.InstallREST --> installer.Install --> a.registerResourceHandlers
// k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/endpoints/installer.go:181
func (a *APIInstaller) registerResourceHandlers(path string, storage rest.Storage, ws *restful.WebService) (*metav1.APIResource, error) {
    ...
    // 一、判斷該 resource 實現了哪些 REST 操做接口，以此來判斷其支持的 verbs 以便爲其添加路由
    // what verbs are supported by the storage, used to know what verbs we support per path
    creater, isCreater := storage.(rest.Creater)
    namedCreater, isNamedCreater := storage.(rest.NamedCreater)
    lister, isLister := storage.(rest.Lister)
    getter, isGetter := storage.(rest.Getter)
    ...
    // 二、爲 resource 添加對應的 actions(+根據是否支持 namespace)
    // Get the list of actions for the given scope.
    switch {
    case !namespaceScoped:
        // Handle non-namespace scoped resources like nodes.
        resourcePath := resource
        resourceParams := params
        itemPath := resourcePath + "/{name}"
        nameParams := append(params, nameParam)
        proxyParams := append(nameParams, pathParam)
        ...
        // Handler for standard REST verbs (GET, PUT, POST and DELETE).
        // Add actions at the resource path: /api/apiVersion/resource
        actions = appendIf(actions, action{"LIST", resourcePath, resourceParams, namer, false}, isLister)
        actions = appendIf(actions, action{"POST", resourcePath, resourceParams, namer, false}, isCreater)
        ...
    }
    ...
    // 三、從 rest.Storage 到 restful.Route 映射
    // 爲每一個操做添加對應的 handler
    for _, action := range actions {
        ...
        switch action.Verb {
        ...
        case "POST": // Create a resource.
            var handler restful.RouteFunction
            // 四、初始化 handler
            if isNamedCreater {
                handler = restfulCreateNamedResource(namedCreater, reqScope, admit)
            } else {
                handler = restfulCreateResource(creater, reqScope, admit)
            }
            handler = metrics.InstrumentRouteFunc(action.Verb, group, version, resource, subresource, requestScope, metrics.APIServerComponent, handler)
            ...
            // 五、route 與 handler 進行綁定    
            route := ws.POST(action.Path).To(handler).
                Doc(doc).
                Param(ws.QueryParameter("pretty", "If 'true', then the output is pretty printed.")).
                Operation("create"+namespaced+kind+strings.Title(subresource)+operationSuffix).
                Produces(append(storageMeta.ProducesMIMETypes(action.Verb), mediaTypes...)...).
                Returns(http.StatusOK, "OK", producedObject).
                // TODO: in some cases, the API may return a v1.Status instead of the versioned object
                // but currently go-restful can't handle multiple different objects being returned.
                Returns(http.StatusCreated, "Created", producedObject).
                Returns(http.StatusAccepted, "Accepted", producedObject).
                Reads(defaultVersionedObject).
                Writes(producedObject)
            if err := AddObjectParams(ws, route, versionedCreateOptions); err != nil {
                return nil, err
            }
            addParams(route, action.Params)
            // 六、添加到路由中    
            routes = append(routes, route)
        case "DELETE": // Delete a resource.
        ...
        default:
            return nil, fmt.Errorf("unrecognized action verb: %s", action.Verb)
        }
        for _, route := range routes {
            route.Metadata(ROUTE_META_GVK, metav1.GroupVersionKind{
                Group:   reqScope.Kind.Group,
                Version: reqScope.Kind.Version,
                Kind:    reqScope.Kind.Kind,
            })
            route.Metadata(ROUTE_META_ACTION, strings.ToLower(action.Verb))
            ws.Route(route)
        }
        // Note: update GetAuthorizerAttributes() when adding a custom handler.
    }
    ...
}

kubeAPIServer代碼結構整理以下：

1. apiserver總體啓動邏輯 k8s.io/kubernetes/cmd/kube-apiserver
2. apiserver bootstrap-controller建立&運行邏輯 k8s.io/kubernetes/pkg/master
3. API Resource對應後端RESTStorage(based on genericregistry.Store)建立k8s.io/kubernetes/pkg/registry
4. aggregated-apiserver建立&處理邏輯 k8s.io/kubernetes/staging/src/k8s.io/kube-aggregator
5. extensions-apiserver建立&處理邏輯 k8s.io/kubernetes/staging/src/k8s.io/apiextensions-apiserver
6. apiserver建立&運行 k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/server
7. 註冊API Resource資源處理handler(InstallREST&Install®isterResourceHandlers) k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/endpoints
8. 建立存儲後端(etcdv3) k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/storage
9. genericregistry.Store.CompleteWithOptions初始化 k8s.io/kubernetes/staging/src/k8s.io/apiserver/pkg/registry

調用鏈整理以下：

更多代碼原理詳情，參考 kubernetes-reading-notes 。

aggregatorServer

aggregatorServer主要用於處理擴展Kubernetes API Resources的第二種方式Aggregated APIServer(AA)，將CR請求代理給AA：

這裏結合Kubernetes官方給出的aggregated apiserver例子sample-apiserver，總結原理以下：

aggregatorServer經過APIServices對象關聯到某個Service來進行請求的轉發，其關聯的Service類型進一步決定了請求轉發的形式。aggregatorServer包括一個GenericAPIServer和維護自身狀態的Controller。其中GenericAPIServer主要處理apiregistration.k8s.io組下的APIService資源請求，而Controller包括：
- apiserviceRegistrationController：負責根據APIService定義的aggregated server service構建代理，將CR的請求轉發給後端的aggregated server
- availableConditionController：維護 APIServices 的可用狀態，包括其引用 Service 是否可用等；
- autoRegistrationController：用於保持 API 中存在的一組特定的 APIServices；
- crdRegistrationController：負責將 CRD GroupVersions 自動註冊到 APIServices 中；
- openAPIAggregationController：將 APIServices 資源的變化同步至提供的 OpenAPI 文檔；

apiserviceRegistrationController負責根據APIService定義的aggregated server service構建代理，將CR的請求轉發給後端的aggregated server。apiService有兩種類型：Local(Service爲空)以及Service(Service非空)。apiserviceRegistrationController負責對這兩種類型apiService設置代理：Local類型會直接路由給kube-apiserver進行處理；而Service類型則會設置代理並將請求轉化爲對aggregated Service的請求(proxyPath := "/apis/" + apiService.Spec.Group + "/" + apiService.Spec.Version)，而請求的負載均衡策略則是優先本地訪問kube-apiserver(若是service爲kubernetes default apiserver service:443)=>經過service ClusterIP:Port訪問(默認) 或者經過隨機選擇service endpoint backend進行訪問：

func (s *APIAggregator) AddAPIService(apiService *v1.APIService) error {
  ...
    proxyPath := "/apis/" + apiService.Spec.Group + "/" + apiService.Spec.Version
    // v1. is a special case for the legacy API.  It proxies to a wider set of endpoints.
    if apiService.Name == legacyAPIServiceName {
        proxyPath = "/api"
    }
    // register the proxy handler
    proxyHandler := &proxyHandler{
        localDelegate:   s.delegateHandler,
        proxyClientCert: s.proxyClientCert,
        proxyClientKey:  s.proxyClientKey,
        proxyTransport:  s.proxyTransport,
        serviceResolver: s.serviceResolver,
        egressSelector:  s.egressSelector,
    }
  ...
    s.proxyHandlers[apiService.Name] = proxyHandler
    s.GenericAPIServer.Handler.NonGoRestfulMux.Handle(proxyPath, proxyHandler)
    s.GenericAPIServer.Handler.NonGoRestfulMux.UnlistedHandlePrefix(proxyPath+"/", proxyHandler)
  ...
    // it's time to register the group aggregation endpoint
    groupPath := "/apis/" + apiService.Spec.Group
    groupDiscoveryHandler := &apiGroupHandler{
        codecs:    aggregatorscheme.Codecs,
        groupName: apiService.Spec.Group,
        lister:    s.lister,
        delegate:  s.delegateHandler,
    }
    // aggregation is protected
    s.GenericAPIServer.Handler.NonGoRestfulMux.Handle(groupPath, groupDiscoveryHandler)
    s.GenericAPIServer.Handler.NonGoRestfulMux.UnlistedHandle(groupPath+"/", groupDiscoveryHandler)
    s.handledGroups.Insert(apiService.Spec.Group)
    return nil
}
// k8s.io/kubernetes/staging/src/k8s.io/kube-aggregator/pkg/apiserver/handler_proxy.go:109
func (r *proxyHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) {
    // 加載roxyHandlingInfo處理請求  
    value := r.handlingInfo.Load()
    if value == nil {
        r.localDelegate.ServeHTTP(w, req)
        return
    }
    handlingInfo := value.(proxyHandlingInfo)
  ...
    // 判斷APIService服務是否正常
    if !handlingInfo.serviceAvailable {
        proxyError(w, req, "service unavailable", http.StatusServiceUnavailable)
        return
    }
    // 將原始請求轉化爲對APIService的請求
    // write a new location based on the existing request pointed at the target service
    location := &url.URL{}
    location.Scheme = "https"
    rloc, err := r.serviceResolver.ResolveEndpoint(handlingInfo.serviceNamespace, handlingInfo.serviceName, handlingInfo.servicePort)
    if err != nil {
        klog.Errorf("error resolving %s/%s: %v", handlingInfo.serviceNamespace, handlingInfo.serviceName, err)
        proxyError(w, req, "service unavailable", http.StatusServiceUnavailable)
        return
    }
    location.Host = rloc.Host
    location.Path = req.URL.Path
    location.RawQuery = req.URL.Query().Encode()
    newReq, cancelFn := newRequestForProxy(location, req)
    defer cancelFn()
   ...
    proxyRoundTripper = transport.NewAuthProxyRoundTripper(user.GetName(), user.GetGroups(), user.GetExtra(), proxyRoundTripper)
    handler := proxy.NewUpgradeAwareHandler(location, proxyRoundTripper, true, upgrade, &responder{w: w})
    handler.ServeHTTP(w, newReq)
}

$ kubectl get APIService           
NAME                                   SERVICE                      AVAILABLE   AGE
...
v1.apps                                Local                        True        50d
...
v1beta1.metrics.k8s.io                 kube-system/metrics-server   True        50d
...

# default APIServices
$ kubectl get -o yaml APIService/v1.apps
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  labels:
    kube-aggregator.kubernetes.io/automanaged: onstart
  name: v1.apps
  selfLink: /apis/apiregistration.k8s.io/v1/apiservices/v1.apps
spec:
  group: apps
  groupPriorityMinimum: 17800
  version: v1
  versionPriority: 15
status:
  conditions:
  - lastTransitionTime: "2020-10-20T10:39:48Z"
    message: Local APIServices are always available
    reason: Local
    status: "True"
    type: Available

# aggregated server    
$ kubectl get -o yaml APIService/v1beta1.metrics.k8s.io
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  labels:
    addonmanager.kubernetes.io/mode: Reconcile
    kubernetes.io/cluster-service: "true"
  name: v1beta1.metrics.k8s.io
  selfLink: /apis/apiregistration.k8s.io/v1/apiservices/v1beta1.metrics.k8s.io
spec:
  group: metrics.k8s.io
  groupPriorityMinimum: 100
  insecureSkipTLSVerify: true
  service:
    name: metrics-server
    namespace: kube-system
    port: 443
  version: v1beta1
  versionPriority: 100
status:
  conditions:
  - lastTransitionTime: "2020-12-05T00:50:48Z"
    message: all checks passed
    reason: Passed
    status: "True"
    type: Available

# CRD
$ kubectl get -o yaml APIService/v1.duyanghao.example.com
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  labels:
    kube-aggregator.kubernetes.io/automanaged: "true"
  name: v1.duyanghao.example.com
  selfLink: /apis/apiregistration.k8s.io/v1/apiservices/v1.duyanghao.example.com
spec:
  group: duyanghao.example.com
  groupPriorityMinimum: 1000
  version: v1
  versionPriority: 100
status:
  conditions:
  - lastTransitionTime: "2020-12-11T08:45:37Z"
    message: Local APIServices are always available
    reason: Local
    status: "True"
    type: Available

aggregatorServer建立過程當中會根據全部kube-apiserver定義的API資源建立默認的APIService列表，名稱便是$VERSION.$GROUP，這些APIService都會有標籤kube-aggregator.kubernetes.io/automanaged: onstart，例如：v1.apps apiService。autoRegistrationController建立並維護這些列表中的APIService，也即咱們看到的Local apiService；對於自定義的APIService(aggregated server)，則不會對其進行處理
aggregated server實現CR(自定義API資源) 的CRUD API接口，並能夠靈活選擇後端存儲，能夠與core kube-apiserver一塊兒公用etcd，也可本身獨立部署etcd數據庫或者其它數據庫。aggregated server實現的CR API路徑爲：/apis/$GROUP/$VERSION，具體到sample apiserver爲：/apis/wardle.example.com/v1alpha1，下面的資源類型有：flunders以及fischers
aggregated server經過部署APIService類型資源，service fields指向對應的aggregated server service實現與core kube-apiserver的集成與交互

sample-apiserver目錄結構以下，可參考編寫本身的aggregated server：

staging/src/k8s.io/sample-apiserver
├── artifacts
│   ├── example
│   │   ├── apiservice.yaml
      ...
├── hack
├── main.go
└── pkg
├── admission
├── apis
├── apiserver
├── cmd
├── generated
│   ├── clientset
│   │   └── versioned
              ...
│   │       └── typed
│   │           └── wardle
│   │               ├── v1alpha1
│   │               └── v1beta1
│   ├── informers
│   │   └── externalversions
│   │       └── wardle
│   │           ├── v1alpha1
│   │           └── v1beta1
│   ├── listers
│   │   └── wardle
│   │       ├── v1alpha1
│   │       └── v1beta1
└── registry

其中，artifacts用於部署yaml示例
hack目錄存放自動腳本(eg: update-codegen)
main.go是aggregated server啓動入口；pkg/cmd負責啓動aggregated server具體邏輯；pkg/apiserver用於aggregated server初始化以及路由註冊
pkg/apis負責相關CR的結構體定義，自動生成(update-codegen)
pkg/admission負責准入的相關代碼
pkg/generated負責生成訪問CR的clientset，informers，以及listers
pkg/registry目錄負責CR相關的RESTStorage實現

更多代碼原理詳情，參考 kubernetes-reading-notes 。

apiExtensionsServer

apiExtensionsServer主要負責CustomResourceDefinition（CRD）apiResources以及apiVersions的註冊，同時處理CRD以及相應CustomResource（CR）的REST請求(若是對應CR不能被處理的話則會返回404)，也是apiserver Delegation的最後一環

原理總結以下：

Custom Resource，簡稱CR，是Kubernetes自定義資源類型，與之相對應的就是Kubernetes內置的各類資源類型，例如Pod、Service等。利用CR咱們能夠定義任何想要的資源類型
CRD經過yaml文件的形式向Kubernetes註冊CR實現自定義api-resources，屬於第二種擴展Kubernetes API資源的方式，也是廣泛使用的一種
APIExtensionServer負責CustomResourceDefinition（CRD）apiResources以及apiVersions的註冊，同時處理CRD以及相應CustomResource（CR）的REST請求(若是對應CR不能被處理的話則會返回404)，也是apiserver Delegation的最後一環
crdRegistrationController負責將CRD GroupVersions自動註冊到APIServices中。具體邏輯爲：枚舉全部CRDs，而後根據CRD定義的crd.Spec.Group以及crd.Spec.Versions字段構建APIService，並添加到autoRegisterController.apiServicesToSync中，由autoRegisterController進行建立以及維護操做。這也是爲何建立完CRD後會產生對應的APIService對象

APIExtensionServer包含的controller以及功能以下所示：

openapiController：將 crd 資源的變化同步至提供的 OpenAPI 文檔，可經過訪問 /openapi/v2 進行查看；
crdController：負責將 crd 信息註冊到 apiVersions 和 apiResources 中，二者的信息可經過 kubectl api-versions 和 kubectl api-resources 查看；

kubectl api-versions命令返回全部Kubernetes集羣資源的版本信息（實際發出了兩個請求，分別是https://127.0.0.1:6443/api以及https://127.0.0.1:6443/apis，並在最後將兩個請求的返回結果進行了合併）

$ kubectl -v=8 api-versions 
I1211 11:44:50.276446   22493 loader.go:375] Config loaded from file:  /root/.kube/config
I1211 11:44:50.277005   22493 round_trippers.go:420] GET https://127.0.0.1:6443/api?timeout=32s
...
I1211 11:44:50.290265   22493 request.go:1068] Response Body: {"kind":"APIVersions","versions":["v1"],"serverAddressByClientCIDRs":[{"clientCIDR":"0.0.0.0/0","serverAddress":"x.x.x.x:6443"}]}
I1211 11:44:50.293673   22493 round_trippers.go:420] GET https://127.0.0.1:6443/apis?timeout=32s
...
I1211 11:44:50.298360   22493 request.go:1068] Response Body: {"kind":"APIGroupList","apiVersion":"v1","groups":[{"name":"apiregistration.k8s.io","versions":[{"groupVersion":"apiregistration.k8s.io/v1","version":"v1"},{"groupVersion":"apiregistration.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"apiregistration.k8s.io/v1","version":"v1"}},{"name":"extensions","versions":[{"groupVersion":"extensions/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"extensions/v1beta1","version":"v1beta1"}},{"name":"apps","versions":[{"groupVersion":"apps/v1","version":"v1"}],"preferredVersion":{"groupVersion":"apps/v1","version":"v1"}},{"name":"events.k8s.io","versions":[{"groupVersion":"events.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"events.k8s.io/v1beta1","version":"v1beta1"}},{"name":"authentication.k8s.io","versions":[{"groupVersion":"authentication.k8s.io/v1","version":"v1"},{"groupVersion":"authentication.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"authentication.k8s.io/v1"," [truncated 4985 chars]
apiextensions.k8s.io/v1
apiextensions.k8s.io/v1beta1
apiregistration.k8s.io/v1
apiregistration.k8s.io/v1beta1
apps/v1
authentication.k8s.io/v1beta1
...
storage.k8s.io/v1
storage.k8s.io/v1beta1
v1

kubectl api-resources命令就是先獲取全部API版本信息，而後對每個API版本調用接口獲取該版本下的全部API資源類型

$ kubectl -v=8 api-resources
 5077 loader.go:375] Config loaded from file:  /root/.kube/config
 I1211 15:19:47.593450   15077 round_trippers.go:420] GET https://127.0.0.1:6443/api?timeout=32s
 I1211 15:19:47.602273   15077 request.go:1068] Response Body: {"kind":"APIVersions","versions":["v1"],"serverAddressByClientCIDRs":[{"clientCIDR":"0.0.0.0/0","serverAddress":"x.x.x.x:6443"}]}
 I1211 15:19:47.606279   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis?timeout=32s
 I1211 15:19:47.610333   15077 request.go:1068] Response Body: {"kind":"APIGroupList","apiVersion":"v1","groups":[{"name":"apiregistration.k8s.io","versions":[{"groupVersion":"apiregistration.k8s.io/v1","version":"v1"},{"groupVersion":"apiregistration.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"apiregistration.k8s.io/v1","version":"v1"}},{"name":"extensions","versions":[{"groupVersion":"extensions/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"extensions/v1beta1","version":"v1beta1"}},{"name":"apps","versions":[{"groupVersion":"apps/v1","version":"v1"}],"preferredVersion":{"groupVersion":"apps/v1","version":"v1"}},{"name":"events.k8s.io","versions":[{"groupVersion":"events.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"events.k8s.io/v1beta1","version":"v1beta1"}},{"name":"authentication.k8s.io","versions":[{"groupVersion":"authentication.k8s.io/v1","version":"v1"},{"groupVersion":"authentication.k8s.io/v1beta1","version":"v1beta1"}],"preferredVersion":{"groupVersion":"authentication.k8s.io/v1"," [truncated 4985 chars]
 I1211 15:19:47.614700   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis/batch/v1?timeout=32s
 I1211 15:19:47.614804   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis/authentication.k8s.io/v1?timeout=32s
 I1211 15:19:47.615687   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis/auth.tkestack.io/v1?timeout=32s
 https://127.0.0.1:6443/apis/authentication.k8s.io/v1beta1?timeout=32s
 I1211 15:19:47.616794   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis/coordination.k8s.io/v1?timeout=32s
 I1211 15:19:47.616863   15077 round_trippers.go:420] GET https://127.0.0.1:6443/apis/apps/v1?timeout=32s
 ...
 NAME                              SHORTNAMES   APIGROUP                       NAMESPACED   KIND
 bindings                                                                      true         Binding
 endpoints                         ep                                          true         Endpoints
 events                            ev                                          true         Event
 limitranges                       limits                                      true         LimitRange
 namespaces                        ns                                          false        Namespace
 nodes                             no                                          false        Node
 ...

namingController：檢查 crd obj 中是否有命名衝突，可在 crd .status.conditions 中查看；
establishingController：檢查 crd 是否處於正常狀態，可在 crd .status.conditions 中查看；
nonStructuralSchemaController：檢查 crd obj 結構是否正常，可在 crd .status.conditions 中查看；
apiApprovalController：檢查 crd 是否遵循 Kubernetes API 聲明策略，可在 crd .status.conditions 中查看；
finalizingController：相似於 finalizes 的功能，與 CRs 的刪除有關；

總結CR CRUD APIServer處理邏輯以下：

createAPIExtensionsServer=>NewCustomResourceDefinitionHandler=>crdHandler=>註冊CR CRUD API接口：

// New returns a new instance of CustomResourceDefinitions from the given config.
func (c completedConfig) New(delegationTarget genericapiserver.DelegationTarget) (*CustomResourceDefinitions, error) {
  ...
    crdHandler, err := NewCustomResourceDefinitionHandler(
      versionDiscoveryHandler,
        groupDiscoveryHandler,
      s.Informers.Apiextensions().V1().CustomResourceDefinitions(),
        delegateHandler,
      c.ExtraConfig.CRDRESTOptionsGetter,
        c.GenericConfig.AdmissionControl,
      establishingController,
        c.ExtraConfig.ServiceResolver,
      c.ExtraConfig.AuthResolverWrapper,
        c.ExtraConfig.MasterCount,
        s.GenericAPIServer.Authorizer,
        c.GenericConfig.RequestTimeout,
        time.Duration(c.GenericConfig.MinRequestTimeout)*time.Second,
        apiGroupInfo.StaticOpenAPISpec,
        c.GenericConfig.MaxRequestBodyBytes,
    )
    if err != nil {
        return nil, err
    }
    s.GenericAPIServer.Handler.NonGoRestfulMux.Handle("/apis", crdHandler)
    s.GenericAPIServer.Handler.NonGoRestfulMux.HandlePrefix("/apis/", crdHandler)
    ...
    return s, nil
}

crdHandler處理邏輯以下：

解析req(GET /apis/duyanghao.example.com/v1/namespaces/default/students)，根據請求路徑中的group(duyanghao.example.com)，version(v1)，以及resource字段(students)獲取對應CRD內容(crd, err := r.crdLister.Get(crdName))
經過crd.UID以及crd.Name獲取crdInfo，若不存在則建立對應的crdInfo(crdInfo, err := r.getOrCreateServingInfoFor(crd.UID, crd.Name))。crdInfo中包含了CRD定義以及該CRD對應Custom Resource的customresource.REST storage
customresource.REST storage由CR對應的Group(duyanghao.example.com)，Version(v1)，Kind(Student)，Resource(students)等建立完成，因爲CR在Kubernetes代碼中並無具體結構體定義，因此這裏會先初始化一個範型結構體Unstructured(用於保存全部類型的Custom Resource)，並對該結構體進行SetGroupVersionKind操做(設置具體Custom Resource Type)

從customresource.REST storage獲取Unstructured結構體後會對其進行相應轉換而後返回

// k8s.io/kubernetes/staging/src/k8s.io/apiextensions-apiserver/pkg/apiserver/customresource_handler.go:223
func (r *crdHandler) ServeHTTP(w http.ResponseWriter, req *http.Request) {
  ctx := req.Context()
  requestInfo, ok := apirequest.RequestInfoFrom(ctx)
  ...
  crdName := requestInfo.Resource + "." + requestInfo.APIGroup
  crd, err := r.crdLister.Get(crdName)
  ...
  crdInfo, err := r.getOrCreateServingInfoFor(crd.UID, crd.Name)
  verb := strings.ToUpper(requestInfo.Verb)
  resource := requestInfo.Resource
  subresource := requestInfo.Subresource
  scope := metrics.CleanScope(requestInfo)
  ...
  switch {
  case subresource == "status" && subresources != nil && subresources.Status != nil:
      handlerFunc = r.serveStatus(w, req, requestInfo, crdInfo, terminating, supportedTypes)
  case subresource == "scale" && subresources != nil && subresources.Scale != nil:
      handlerFunc = r.serveScale(w, req, requestInfo, crdInfo, terminating, supportedTypes)
  case len(subresource) == 0:
      handlerFunc = r.serveResource(w, req, requestInfo, crdInfo, terminating, supportedTypes)
  default:
      responsewriters.ErrorNegotiated(
          apierrors.NewNotFound(schema.GroupResource{Group: requestInfo.APIGroup, Resource: requestInfo.Resource}, requestInfo.Name),
          Codecs, schema.GroupVersion{Group: requestInfo.APIGroup, Version: requestInfo.APIVersion}, w, req,
      )
  }
  if handlerFunc != nil {
      handlerFunc = metrics.InstrumentHandlerFunc(verb, requestInfo.APIGroup, requestInfo.APIVersion, resource, subresource, scope, metrics.APIServerComponent, handlerFunc)
      handler := genericfilters.WithWaitGroup(handlerFunc, longRunningFilter, crdInfo.waitGroup)
      handler.ServeHTTP(w, req)
      return
  }
}

更多代碼原理詳情，參考 kubernetes-reading-notes 。

Conclusion

本文從源碼層面對Kubernetes apiserver進行了一個概覽性總結，包括：aggregatorServer，kubeAPIServer，apiExtensionsServer以及bootstrap-controller等。經過閱讀本文能夠對apiserver內部原理有一個大體的理解，另外也有助於後續深刻研究