K8S中各組件和Kube Apiserver通信時的認證和鑒權 - 網安 - 專業的網絡安全產業、社區、知識平臺

背景

和master節點kube api-server通信的組件有很多，包括：

kubelet
calico
scheduler
kubectl
某些pod可能會和kube api-server通信

這些組件和api-server通信時用的是什么身份，可以操作哪些api資源呢？

本文使用的k8s集群是用kubekey^[1]搭建，命令是./kk create cluster --with-kubernetes v1.21.5 --with-kubesphere v3.2.1

Kubectl的身份和權限

Kubectl用的是什么身份？

Kubectl默認會用到.kube/config配置，其中包含證書信息。

  root@ip-172-31-14-204:~# cat .kube/config  apiVersion: v1  ...  users:  - name: kubernetes-admin    user:      client-certificate-data: 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  ...

可以看到這個證書的CN是kubernetes-admin，O是system:masters。根據文檔^[2]可以知道，這表示證書代表的用戶是kubernetes-admin，用戶組是system:masters。

  root@ip-172-31-14-204:~# echo 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 | base64 -d | openssl x509  -noout -text  Certificate:      ...          Subject: O = system:masters, CN = kubernetes-admin

那這個證書代表的用戶能操作哪些資源呢？

能操作哪些資源呢？

這個需要看"用戶kubernetes-admin"、"用戶組system:masters"在集群中綁定了什么"角色"。

可以看到它綁定了"ClusterRole/cluster-admin"，這個角色可以對所有資源做任意操作。

  root@ip-172-31-14-204:~# kubectl get ClusterRoleBinding -A -o wide | grep system:masters  cluster-admin                                          ClusterRole/cluster-admin                                                          33h                                    system:masters

可以通過kubectl get ClusterRole/cluster-admin -o yaml查看角色的權限。

所以，.kube/config證書中代表的用戶身份可以對所有資源做任意操作。

Kube-scheduler的身份和權限

Kube-scheduler用的是什么身份？

在master節點上查看scheduler進程，可以"大膽猜測"用的是/etc/kubernetes/scheduler.conf中的證書信息。

在我的k8s環境中，kube-scheduler是運行在pod中的，不過pod和宿主機的/etc/kubernetes/scheduler.conf文件是一樣的。

  root@ip-172-31-14-33:~# ps aux|grep kube-scheduler  root     51897  0.2  0.7 753012 57736 ?        Ssl  May25   3:59 kube-scheduler --authentication-kubeconfig=/etc/kubernetes/scheduler.conf --authorization-kubeconfig=/etc/kubernetes/scheduler.conf --bind-address=0.0.0.0 --feature-gates=RotateKubeletServerCertificate=true,TTLAfterFinished=true,ExpandCSIVolumes=true,CSIStorageCapacity=true --kubeconfig=/etc/kubernetes/scheduler.conf --leader-elect=true --port=0

查看證書的Subject信息，可以看到用戶是system:kube-scheduler

  root@ip-172-31-14-33:~# echo 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 | base64 -d |openssl x509  -noout -text  Certificate:      ...          Subject: CN = system:kube-scheduler

可以看到"用戶system:kube-scheduler"綁定了兩個角色。

  root@ip-172-31-14-33:~# kubectl get ClusterRoleBindings -o wide |grep system:kube-scheduler  system:kube-scheduler                                  ClusterRole/system:kube-scheduler                                                  31h   system:kube-scheduler  system:volume-scheduler                                ClusterRole/system:volume-scheduler                                                31h   system:kube-scheduler

Kubelet的身份和權限

Kubelet用的是什么身份？

  root@ip-172-31-14-204:~# ps aux|grep kubelet  root     54396  7.7  2.8 1966128 113860 ?      Ssl  08:14  29:54 /usr/local/bin/kubelet ... --kubeconfig=/etc/kubernetes/kubelet.conf ...

在worker節點上可以看到kubelet進程參數，api-server信息在/etc/kubernetes/kubelet.conf文件中

可以看到是通過證書認證的

  root@ip-172-31-14-204:/etc/kubernetes/manifests# cat /etc/kubernetes/kubelet.conf  apiVersion: v1  ...  users:  - name: system:node:ip-172-31-14-33    user:      client-certificate: /var/lib/kubelet/pki/kubelet-client-current.pem      client-key: /var/lib/kubelet/pki/kubelet-client-current.pem

查看證書的Subject信息，可以知道證書代表的"用戶"是"system:node:ip-172-31-14-204"，"用戶組"是"system:nodes"

  root@ip-172-31-14-204:~# openssl x509 -in /var/lib/kubelet/pki/kubelet-client-current.pem -noout -text  Certificate:      ...          Subject: O = system:nodes, CN = system:node:ip-172-31-14-204

那這個kubelet的證書代表的用戶能操作哪些資源呢？它可以用來像"集群管理員"那樣創建pod嗎？

kubelet能操作哪些資源？

按照之前流程，我們來看一下"用戶system:node:ip-172-31-14-204"、"用戶組system:nodes"在集群中綁定了什么"角色"。

 root@ip-172-31-14-204:~# kubectl get ClusterRoleBinding -A|grep system:nodes  root@ip-172-31-14-204:~# kubectl get ClusterRoleBinding -A|grep "system:node:ip-172-31-14-204"  root@ip-172-31-14-204:~#

會發現"用戶和用戶組"沒有綁定任何一個角色，這和kubectl、kube-scheduler就很不一樣了。

在使用Node鑒權^[3] 文檔中提到，kube apiserver對kubelet的鑒權比較特殊。

當發現請求的用戶在system:nodes組中，用戶名是system:node:時，就限制這個請求只能做有限的操作，比如

讀操作

services
pod
綁定到當前node的pod的secret、configMap

寫操作（如果開啟了NodeRestriction準入插件，就只能修改kubelet所在node的資源）

創建節點、修改節點狀態
創建pod、pod狀態

似乎如果沒有開啟NodeRestriction準入插件，就能讓kubelet在任意node上創建pod。

紅藍對抗中，如果能讓kubelet在任意node上創建pod，就能用來橫移。

文檔中寫到，這里的能力可能隨著k8s版本變化而變化，以確保kubelet最小權限，默認安全。所以，kubelet到底能操作哪些資源，感覺還是來測試一下比較好，下面就來驗證一下kubelet的權限。

驗證kubelet的權限

我們先用kubelet配置文件覆蓋kubectl的默認配置，然后就可以用kubectl命令來驗證。

  root@ip-172-31-14-204:~# mv .kube/config .kube/config.bak  root@ip-172-31-14-204:~# ps aux|grep kubele  root     59075  7.9  2.7 1956332 108712 ?      Ssl  03:37   2:51 /usr/local/bin/kubelet ... --kubeconfig=/etc/kubernetes/kubelet.conf ...  root@ip-172-31-14-204:~# cp /etc/kubernetes/kubelet.conf .kube/config

可以看到不能夠創建pod

 root@ip-172-31-14-204:~# kubectl run httpbin --image kennethreitz/httpbin  Error from server (Forbidden): pods "httpbin" is forbidden: pod does not have "kubernetes.io/config.mirror" annotation, node "ip-172-31-14-204" can only create mirror pod

查看pod信息是可以的

  root@ip-172-31-14-204:~# kubectl get pods -A  NAMESPACE                      NAME                                               READY   STATUS      RESTARTS   AGE  default                        tail                                               1/1     Running     0          25h  kube-system                    calico-kube-controllers-846b5f484d-vzlhw           1/1     Running     0          27h  ...  root@ip-172-31-14-204:~# kubectl describe pod devops-27558900-m69pc -n kubesphere-devops-system  Name:         devops-27558900-m69pc  ...

查看secret和configMap是不允許的

  root@ip-172-31-14-204:~# kubectl get secrets -A  Error from server (Forbidden): secrets is forbidden: User "system:node:ip-172-31-14-204" cannot list resource "secrets" in API group "" at the cluster scope: can only read namespaced object of this type  root@ip-172-31-14-204:~# kubectl get configmap  Error from server (Forbidden): configmaps is forbidden: User "system:node:ip-172-31-14-204" cannot list resource "configmaps" in API group "" in the namespace "default": No Object name found  root@ip-172-31-14-204:~#

小結：kubelet證書可以用來查看pod信息，不能創建pod、不能查看所有命名空間的secret和configMap。看起來和文檔中的說明一致。

calico

calico用的是什么身份？

 root@ip-172-31-14-204:~# cat /etc/cni/net.d/calico-kubeconfig  # Kubeconfig file for Calico CNI plugin. Installed by calico/node.  ...  users:  - name: calico    user:      token: eyJhbGciOiJSUzI1NiIsImtpZCI6ImV6bEstc2QtRTNTaE1ySFg2SWdxMjY2aHpkVFBEWU56SGdfaVdZRXZ4YncifQ.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.mF1Yek_sSBa2nlQIbdJlEZtv3anjIUFPFpj8Ta8Zn1t6vxYEZjswbPxrw-90sbEVGR30GWytUjr3X1tmjWTP0fL9ltAPvcM0tRg6MoLuGdC2uZFbt-zfKFEn42Yme-NuVOKO3K2otgZNt0ym0gYRT41wLCqOuKLq-SHAkHdvIOZ1eVAu0OsN3ccgSFf2nsuBqbDphd4ShqRPNXl1m66UEDlb4WiPmuSPuzRmMiV706YbBPxMUzn9Ve4o4IfBbeBufp_edSWPPW763EGqZmI7qZg-78SUbtAJ6m8Qkq6TrDIcaJbI3Mrl4EsrnE3v3MGMOogfXXs3-yU-NZl3ilQt6Q  ...

可以看到用的是token令牌。這個token令牌是一個jwt字符串，base64解碼后的payload部分如下，可以看到"ServiceAccount"是calico-node，命名空間是kube-system

  {"aud":["https://kubernetes.default.svc.cluster.local"],"exp":1685022900,"iat":1653486900,"iss":"https://kubernetes.default.svc.cluster.local","kubernetes.io":{"namespace":"kube-system","pod":{"name":"calico-node-qdtfb","uid":"87e8fc7d-2a89-4598-9d28-fd624b089603"},"serviceaccount":{"name":"calico-node","uid":"913750eb-bda4-4b2f-8167-46c991699add"},"warnafter":1653490507},"nbf":1653486900,"sub":"system:serviceaccount:kube-system:calico-nodeIn0

可以看到，這個"ServiceAccount"被綁定到了"ClusterRole/calico-node"角色。

  root@ip-172-31-14-204:/home/ubuntu# kubectl get clusterrolebinding -o wide|grep kube-system/calico-node  calico-node                                            ClusterRole/calico-node                                                            34h                                                                                      kube-system/calico-node

pod

pod用的是什么身份？

大部分pod會掛載一個token在/var/run/secrets/kubernetes.io/serviceaccount/token位置，這個token也是一個jwt字符串。

  root@ip-172-31-14-204:/home/ubuntu# kubectl exec -ti tail -- sh  / # cat /var/run/secrets/kubernetes.io/serviceaccount/token  eyJhbGciOiJSUzI1NiIsImtpZCI6ImV6bEstc2QtRTNTaE1ySFg2SWdxMjY2aHpkVFBEWU56SGdfaVdZRXZ4YncifQ.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.L388ZKZWvHYf7Yvt6tA7t8k2QMYcsI9BDcYtoxAgOJiEiwf3LKFQmdH8KF4PnI0kjM3Bg80WztFznotwZqCwCNfXMVMl4_iLGHDAgVB3tsdv9ljZ9FUJbn52PD5aWHSWRqjpyQzv8_89dlnnbGQLHg4M8Ly4OkGuWUOnE1x6vgSa1MkjhrnrJEPnnJo5Fy_vyRdvO2A9iyGh7cC97Ns6WWFkeD7741wSkGkoNkZKqJTyfaa_KScprPiVPYuisi4HkYrP71NzZA_i34Dk-IsomySR4h2WWw_88-kfL_lWZ8PDu5NuVekZZ4xfIQjA6oDhXT_Hx4iIlhwVwgYuTW4V-g

base64解碼后，payload中也能看到一個服務賬號ServiceAccount，這個ServiceAccount也有可能和一個"角色"綁定。你可以動手查看一下自己的pod。

總結

k8s中"X509 客戶證書"和"服務賬號令牌"應該是身份認證最常見的兩種方式，都能代表"用戶"或"用戶組"信息。

大部分的組件(包括scheduler、calico、kubectl、pod)鑒權都是依賴RBAC模型，也就是"用戶或"用戶組"綁定的"角色"規定了能訪問哪些資源。

kubelet的鑒權比較特殊，是依賴Node鑒權模型。

用戶認證^[4]文檔中還提到api訪問控制的其他方法。