RBAC

大家好，我是博哥爱运维，在k8s上我们如何控制访问权限呢，答案就是Role-based access control (RBAC) - 基于角色（Role）的访问控制，（RBAC）是一种基于组织中用户的角色来调节控制对 计算机或网络资源的访问的方法。

在早期的K8s版本，RBAC还未出现的时候，整个K8s的安全是较为薄弱的，有了RBAC后，我们可以对K8s集群的访问人员作非常明细化的控制，控制他们能访问什么资源，以只读还是可以读写的形式来访问，目前RBAC是K8s默认的安全授权标准，所以我们非常有必要来掌握RBAC的使用，这样才有更有力的保障我们的K8s集群的安全使用，下面我们将以生产中的实际使用来大家了解及掌握RBAC的生产应用。

RBAC里面的几种资源关系图，下面将用下面的资源来演示生产中经典的RBAC应用

                  |--- Role --- RoleBinding                只在指定namespace中生效
ServiceAccount ---|
                  |--- ClusterRole --- ClusterRoleBinding  不受namespace限制，在整个K8s集群中生效

在我看来，RBAC在K8s上的用途主要分为两大类：

第一类是保证在K8s上运行的pod服务具有相应的集群权限，如gitlab的CI/CD，它需要能访问除自身以外其他pod，比如gitlab-runner的pod的权限，再比如gitlab-runner的pod需要拥有创建新的临时pod的权限，用以来构建CI/CD自动化流水线，这里大家没用过不懂没关系，先简单了解下就可以了，在本课程后面基于K8s及gitlab的生产实战CI/CD内容会给大家作详细实战讲解；

第二类是创建能访问K8s相应资源、拥有对应权限的kube-config配置给到使用K8s的人员，来作为连接K8s的授权凭证

第一类的实战这里先暂时以早期的helm2来作下讲解，helm是一个快捷安装K8s各类资源的管理工具，通过之前给大家讲解的，一个较为完整的服务可能会存在deployment，service，configmap，secret，ingress等资源来组合使用，大家在用的过程中可能会觉得配置使用较为麻烦，这时候helm就出现了，它把这些资源都打包封装成它自己能识别的内容，我们在安装一个服务的时候，就只需要作下简单的配置，一条命令即可完成上述众多资源的配置安装，titller相当于helm的服务端，它是需要有权限在K8s中创建各类资源的，在初始安装使用时，如果没有配置RBAC权限，我们会看到如下报错：

root@node1:~# helm install stable/mysql
Error: no available release name found

这时，我们可以来快速解决这个问题，创建sa关联K8s自带的最高权限的ClusterRole（生产中建议不要这样做，权限太高有安全隐患，这个就和linux的root管理帐号一样，一般都是建议通过sudo来控制帐号权限）

kubectl create serviceaccount --namespace kube-system tiller
kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
kubectl patch deploy --namespace kube-system tiller-deploy -p '{"spec":{"template":{"spec":{"serviceAccount":"tiller"}}}}'

第二类，我这里就直接以我在生产中实施的完整脚本来做讲解及实战，相信会给大家带来一个全新的学习感受，并能很快掌握它们：

  1. 创建对指定namespace有所有权限的kube-config
#!/bin/bash
#
# This Script based on  https://jeremievallee.com/2018/05/28/kubernetes-rbac-namespace-user.html
# K8s'RBAC doc:         https://kubernetes.io/docs/reference/access-authn-authz/rbac
# Gitlab'CI/CD doc:     hhttps://docs.gitlab.com/ee/user/permissions.html#running-pipelines-on-protected-branches
#
# In honor of the remarkable Windson

BASEDIR="$(dirname "$0")"
folder="$BASEDIR/kube_config"

echo -e "All namespaces is here: \n$(kubectl get ns|awk 'NR!=1{print $1}')"
echo "endpoint server if local network you can use $(kubectl cluster-info |awk '/Kubernetes/{print $NF}')"

namespace=$1
endpoint=$(echo "$2" | sed -e 's,https\?://,,g')

if [[ -z "$endpoint" || -z "$namespace" ]]; then
    echo "Use "$(basename "$0")" NAMESPACE ENDPOINT";
    exit 1;
fi

if ! kubectl get ns|awk 'NR!=1{print $1}'|grep -w "$namespace";then kubectl create ns "$namespace";else echo "namespace: $namespace was exist." ;fi

echo "---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: $namespace-user
  namespace: $namespace
---
kind: Role
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-full-access
  namespace: $namespace
rules:
- apiGroups: ['', 'extensions', 'apps', 'metrics.k8s.io']
  resources: ['*']
  verbs: ['*']
- apiGroups: ['batch']
  resources:
  - jobs
  - cronjobs
  verbs: ['*']
---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-view
  namespace: $namespace
subjects:
- kind: ServiceAccount
  name: $namespace-user
  namespace: $namespace
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: $namespace-user-full-access
---
# https://kubernetes.io/zh/docs/concepts/policy/resource-quotas/
apiVersion: v1
kind: ResourceQuota
metadata:
  name: $namespace-compute-resources
  namespace: $namespace
spec:
  hard:
    pods: "10"
    services: "10"
    persistentvolumeclaims: "5"
    requests.cpu: "1"
    requests.memory: 2Gi
    limits.cpu: "2"
    limits.memory: 4Gi" | kubectl apply -f -
kubectl -n $namespace describe quota $namespace-compute-resources
mkdir -p $folder
tokenName=$(kubectl get sa $namespace-user -n $namespace -o "jsonpath={.secrets[0].name}")
token=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data.token}" | base64 --decode)
certificate=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data['ca\.crt']}")

echo "apiVersion: v1
kind: Config
preferences: {}
clusters:
- cluster:
    certificate-authority-data: $certificate
    server: https://$endpoint
  name: $namespace-cluster
users:
- name: $namespace-user
  user:
    as-user-extra: {}
    client-key-data: $certificate
    token: $token
contexts:
- context:
    cluster: $namespace-cluster
    namespace: $namespace
    user: $namespace-user
  name: $namespace
current-context: $namespace" > $folder/$namespace.kube.conf
  1. 创建对指定namespace有所有权限的kube-config（在已有的namespace中创建）
#!/bin/bash


BASEDIR="$(dirname "$0")"
folder="$BASEDIR/kube_config"

echo -e "All namespaces is here: \n$(kubectl get ns|awk 'NR!=1{print $1}')"
echo "endpoint server if local network you can use $(kubectl cluster-info |awk '/Kubernetes/{print $NF}')"

namespace=$1
endpoint=$(echo "$2" | sed -e 's,https\?://,,g')

if [[ -z "$endpoint" || -z "$namespace" ]]; then
    echo "Use "$(basename "$0")" NAMESPACE ENDPOINT";
    exit 1;
fi


echo "---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: $namespace-user
  namespace: $namespace
---
kind: Role
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-full-access
  namespace: $namespace
rules:
- apiGroups: ['', 'extensions', 'apps', 'metrics.k8s.io']
  resources: ['*']
  verbs: ['*']
- apiGroups: ['batch']
  resources:
  - jobs
  - cronjobs
  verbs: ['*']
---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-view
  namespace: $namespace
subjects:
- kind: ServiceAccount
  name: $namespace-user
  namespace: $namespace
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: $namespace-user-full-access" | kubectl apply -f -

mkdir -p $folder
tokenName=$(kubectl get sa $namespace-user -n $namespace -o "jsonpath={.secrets[0].name}")
token=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data.token}" | base64 --decode)
certificate=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data['ca\.crt']}")

echo "apiVersion: v1
kind: Config
preferences: {}
clusters:
- cluster:
    certificate-authority-data: $certificate
    server: https://$endpoint
  name: $namespace-cluster
users:
- name: $namespace-user
  user:
    as-user-extra: {}
    client-key-data: $certificate
    token: $token
contexts:
- context:
    cluster: $namespace-cluster
    namespace: $namespace
    user: $namespace-user
  name: $namespace
current-context: $namespace" > $folder/$namespace.kube.conf
  1. 同上，创建只读权限的
#!/bin/bash


BASEDIR="$(dirname "$0")"
folder="$BASEDIR/kube_config"

echo -e "All namespaces is here: \n$(kubectl get ns|awk 'NR!=1{print $1}')"
echo "endpoint server if local network you can use $(kubectl cluster-info |awk '/Kubernetes/{print $NF}')"

namespace=$1
endpoint=$(echo "$2" | sed -e 's,https\?://,,g')

if [[ -z "$endpoint" || -z "$namespace" ]]; then
    echo "Use "$(basename "$0")" NAMESPACE ENDPOINT";
    exit 1;
fi


echo "---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: $namespace-user-readonly
  namespace: $namespace
---
kind: Role
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-readonly-access
  namespace: $namespace
rules:
- apiGroups: ['', 'extensions', 'apps', 'metrics.k8s.io']
  resources: ['pods', 'pods/log']
  verbs: ['get', 'list', 'watch']
- apiGroups: ['batch']
  resources: ['jobs', 'cronjobs']
  verbs: ['get', 'list', 'watch']
---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: $namespace-user-view-readonly
  namespace: $namespace
subjects:
- kind: ServiceAccount
  name: $namespace-user-readonly
  namespace: $namespace
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: $namespace-user-readonly-access" | kubectl apply -f -

mkdir -p $folder
tokenName=$(kubectl get sa $namespace-user-readonly -n $namespace -o "jsonpath={.secrets[0].name}")
token=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data.token}" | base64 --decode)
certificate=$(kubectl get secret $tokenName -n $namespace -o "jsonpath={.data['ca\.crt']}")

echo "apiVersion: v1
kind: Config
preferences: {}
clusters:
- cluster:
    certificate-authority-data: $certificate
    server: https://$endpoint
  name: $namespace-cluster-readonly
users:
- name: $namespace-user-readonly
  user:
    as-user-extra: {}
    client-key-data: $certificate
    token: $token
contexts:
- context:
    cluster: $namespace-cluster-readonly
    namespace: $namespace
    user: $namespace-user-readonly
  name: $namespace
current-context: $namespace" > $folder/$namespace-readonly.kube.conf

最后，来一个多个集群配置融合的创建，这个在多集群管理方面非常有用，这里只以创建只读权限配置作为演示

#!/bin/bash
# describe: create k8s cluster all namespaces resources with readonly clusterrole, no exec 、delete ...

# look system default to study:
# kubectl describe clusterrole view

# restore all change:
#kubectl -n kube-system delete sa all-readonly-${clustername}
#kubectl delete clusterrolebinding all-readonly-${clustername}
#kubectl delete clusterrole all-readonly-${clustername}


clustername=$1

Help(){
    echo "Use "$(basename "$0")" ClusterName(example: k8s1|k8s2|k8s3|delk8s1|delk8s2|delk8s3|3in1)";
    exit 1;
}

if [[ -z "${clustername}" ]]; then
    Help
fi

case ${clustername} in
    k8s1)
    endpoint="https://x.x.x.x:123456"
    ;;
    k8s2)
    endpoint="https://x.x.x.x:123456"
    ;;
    k8s3)
    endpoint="https://x.x.x.x:123456"
    ;;
    delk8s1)
    kubectl -n kube-system delete sa all-readonly-k8s1
    kubectl delete clusterrolebinding all-readonly-k8s1
    kubectl delete clusterrole all-readonly-k8s1
    echo "${clustername} successful."
    exit 0
    ;;
    delk8s2)
    kubectl -n kube-system delete sa all-readonly-k8s2
    kubectl delete clusterrolebinding all-readonly-k8s2
    kubectl delete clusterrole all-readonly-k8s2
    echo "${clustername} successful."
    exit 0
    ;;
    delk8s3)
    kubectl -n kube-system delete sa all-readonly-k8s3
    kubectl delete clusterrolebinding all-readonly-k8s3
    kubectl delete clusterrole all-readonly-k8s3
    echo "${clustername} successful."
    exit 0
    ;;
    3in1)
    KUBECONFIG=./all-readonly-k8s1.conf:all-readonly-k8s2.conf:all-readonly-k8s3.conf kubectl config view --flatten > ./all-readonly-3in1.conf
    kubectl --kubeconfig=./all-readonly-3in1.conf config use-context "k8s3"
    kubectl --kubeconfig=./all-readonly-3in1.conf config set-context "k8s3" --namespace="default"
    kubectl --kubeconfig=./all-readonly-3in1.conf config get-contexts
    echo -e "\n\n\n"
    cat ./all-readonly-3in1.conf |base64 -w 0
    exit 0
    ;;
    *)
    Help
esac

echo "---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: all-readonly-${clustername}
rules:
- apiGroups:
  - ''
  resources:
  - configmaps
  - endpoints
  - persistentvolumes
  - persistentvolumeclaims
  - pods
  - replicationcontrollers
  - replicationcontrollers/scale
  - serviceaccounts
  - services
  - nodes
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - ''
  resources:
  - bindings
  - events
  - limitranges
  - namespaces/status
  - pods/log
  - pods/status
  - replicationcontrollers/status
  - resourcequotas
  - resourcequotas/status
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - ''
  resources:
  - namespaces
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - apps
  resources:
  - controllerrevisions
  - daemonsets
  - deployments
  - deployments/scale
  - replicasets
  - replicasets/scale
  - statefulsets
  - statefulsets/scale
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - autoscaling
  resources:
  - horizontalpodautoscalers
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - batch
  resources:
  - cronjobs
  - jobs
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - extensions
  resources:
  - daemonsets
  - deployments
  - deployments/scale
  - ingresses
  - networkpolicies
  - replicasets
  - replicasets/scale
  - replicationcontrollers/scale
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - policy
  resources:
  - poddisruptionbudgets
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - networking.k8s.io
  resources:
  - networkpolicies
  verbs:
  - get
  - list
  - watch
- apiGroups:
  - metrics.k8s.io
  resources:
  - pods
  verbs:
  - get
  - list
  - watch" | kubectl apply -f -

kubectl -n kube-system create sa all-readonly-${clustername}
kubectl create clusterrolebinding all-readonly-${clustername} --clusterrole=all-readonly-${clustername} --serviceaccount=kube-system:all-readonly-${clustername}

tokenName=$(kubectl -n kube-system get sa all-readonly-${clustername} -o "jsonpath={.secrets[0].name}")
token=$(kubectl -n kube-system get secret $tokenName -o "jsonpath={.data.token}" | base64 --decode)
certificate=$(kubectl -n kube-system get secret $tokenName -o "jsonpath={.data['ca\.crt']}")

echo "apiVersion: v1
kind: Config
preferences: {}
clusters:
- cluster:
    certificate-authority-data: $certificate
    server: $endpoint
  name: all-readonly-${clustername}-cluster
users:
- name: all-readonly-${clustername}
  user:
    as-user-extra: {}
    client-key-data: $certificate
    token: $token
contexts:
- context:
    cluster: all-readonly-${clustername}-cluster
    user: all-readonly-${clustername}
  name: ${clustername}
current-context: ${clustername}" > ./all-readonly-${clustername}.conf
文档更新时间: 2021-07-28 16:55   作者：李延召