前言:
kubernetes其实也需要有一定的安全,权限外溢会导致整个系统的破坏,比如,被人恶意种挖矿木马,或者遭遇勒索病毒,因此,在进行kubernetes集群的管理工作时,我们应该给账号划分多层次的账号从而满足各种各样的需求。
一,
serviceaccount形式的账号,此账号只有查看各类资源的功能,没有操作资源的功能
(1)
建立一个namespace,命名为view,建立一个sa名字为user1-view
[root@master ~]# k create ns view
namespace/view created
[root@master ~]# k create sa user1-view -n view
serviceaccount/user1-view created
(2)
在集群中有几个默认存在的clusterrole,其中的view这个clusterrole是具有查看所有资源的权限,我们将此clusterrole绑定到user-view上,那么,在日常的使用中,只需要登录这个sa就可以方便的查看所有的资源了,但,重要的资源此sa无权删除或更改,从而提高了集群的安全性。
[root@master ~]# cat user1-view-bind.yaml
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: role-bind-user1
namespace: view
subjects:
- kind: ServiceAccount
name: user1-view
namespace: view
roleRef:
kind: ClusterRole
name: view
apiGroup: rbac.authorization.k8s.io
验证:
查询此sa的token,命令如下:
[root@master ~]# k describe sa user1-view -n view
Name: user1-view
Namespace: view
Labels: <none>
Annotations: <none>
Image pull secrets: <none>
Mountable secrets: user1-view-token-7qp6t
Tokens: user1-view-token-7qp6t
Events: <none>
[root@master ~]# k describe secret user1-view-token-7qp6t -n view
Name: user1-view-token-7qp6t
Namespace: view
Labels: <none>
Annotations: kubernetes.io/service-account.name: user1-view
kubernetes.io/service-account.uid: 9a7d3693-7d11-4254-a3b1-6842ef8fcd8c
Type: kubernetes.io/service-account-token
Data
====
ca.crt: 1359 bytes
namespace: 4 bytes
token: eyJhbGciOiJSUzI1NiIsImtpZCI6IkxNRk93NFRwc3l5UlJjcG05V1IwY25rWjNGOWM1Z05OQjVXN1ROa004R1UifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJ2aWV3Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZWNyZXQubmFtZSI6InVzZXIxLXZpZXctdG9rZW4tN3FwNnQiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC5uYW1lIjoidXNlcjEtdmlldyIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50LnVpZCI6IjlhN2QzNjkzLTdkMTEtNDI1NC1hM2IxLTY4NDJlZjhmY2Q4YyIsInN1YiI6InN5c3RlbTpzZXJ2aWNlYWNjb3VudDp2aWV3OnVzZXIxLXZpZXcifQ.h-9-w-02bzh1ccujta4Q4BOTy-TSZ0AZj6HdNYm_VNVVD7CevRE0VHERN7CzhYgIIhK_6FM5O6P7HjhSr92Lb5QW1jJSecK6_ywK7f_BGczvDrqVcQ0TPpUIl0U7Q_UemB_-tJ8X9s2bvRrl2U-BvOlTBZdRs7sLOGE5GzgwmAIHooQeAhKcRwPUDmxlZ0XChjeUVoREupKJEPJFF8T6zX9LXXm4DAV42Qyu2IzpkoEk2xsm0jYNcoTAj6lOSoMmtL8kSTYrvoWU1dmxIysyuR_pDgyv5f9-49NOW8fbxR10kv3Ii9PtT4O-mxyr81bUuWOLMoD5IoWaet3tpxYMGA
也可以这样查询token:
[root@master ~]# k get secret -n view |grep user1
user1-view-token-7qp6t kubernetes.io/service-account-token 3 73m
[root@master ~]# kubectl get secret user1-view-token-7qp6t -o jsonpath={.data.token} -n view |base64 -d && echo
eyJhbGciOiJSUzI1NiIsImtpZCI6IkxNRk93NFRwc3l5UlJjcG05V1IwY25rWjNGOWM1Z05OQjVXN1ROa004R1UifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJ2aWV3Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZWNyZXQubmFtZSI6InVzZXIxLXZpZXctdG9rZW4tN3FwNnQiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC5uYW1lIjoidXNlcjEtdmlldyIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50LnVpZCI6IjlhN2QzNjkzLTdkMTEtNDI1NC1hM2IxLTY4NDJlZjhmY2Q4YyIsInN1YiI6InN5c3RlbTpzZXJ2aWNlYWNjb3VudDp2aWV3OnVzZXIxLXZpZXcifQ.h-9-w-02bzh1ccujta4Q4BOTy-TSZ0AZj6HdNYm_VNVVD7CevRE0VHERN7CzhYgIIhK_6FM5O6P7HjhSr92Lb5QW1jJSecK6_ywK7f_BGczvDrqVcQ0TPpUIl0U7Q_UemB_-tJ8X9s2bvRrl2U-BvOlTBZdRs7sLOGE5GzgwmAIHooQeAhKcRwPUDmxlZ0XChjeUVoREupKJEPJFF8T6zX9LXXm4DAV42Qyu2IzpkoEk2xsm0jYNcoTAj6lOSoMmtL8kSTYrvoWU1dmxIysyuR_pDgyv5f9-49NOW8fbxR10kv3Ii9PtT4O-mxyr81bUuWOLMoD5IoWaet3tpxYMGA
dashboard登录成功:
随便找个pod看能不能删除(当然是无权删除):
当然了,更改任何配置此sa也是没有权限的,secret这样的敏感信息也是无权查看的哦,这样集群就非常的安全啦。
上述的权限管理是利用了集群内置的角色view实现的,那么这样的权限外放是显得有点粗放的,并不是非常的精细,如果只是希望某个sa只能够访问指定的namespace下资源,如何做呢?
二,权限精细化分配---通过sa和自建角色实现权限精细化分配
(1)
新建一个sa
kubectl create ns dev 先建立一个namespace
[root@master sa]# cat sa-create.yaml
apiVersion: v1
kind: ServiceAccount
metadata:
name: sa-dev
namespace: dev
(2)
建立一个角色,并将该角色绑定到sa上:
角色role-sa 具有的权限仅仅是namespace dev内的所有pod的查看权限,以及deployment的查看权限,无权删除修改这些资源
[root@master sa]# cat sa-role-binding.yaml
kind: Role
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: role-sa
namespace: dev #指定 Namespace
rules: #权限分配
- apiGroups: [""]
resources: ["pods"]
verbs: ["get", "watch", "list"]
- apiGroups: [""]
resources: ["pods/log"]
verbs: ["get","list","watch"]
- apiGroups: [""]
resources: ["pods/attach"]
verbs: ["get","list","watch"]
- apiGroups: [""]
resources: ["pods/exec"]
verbs: ["get","list","watch"]
- apiGroups: [""]
resources: ["pods/status"]
verbs: ["get","list","watch"]
- apiGroups: [""]
resources: ["podtemplates"]
verbs: ["get","list","watch"]
- apiGroups: ["extensions", "apps"]
resources: ["deployments","statefulsets"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["configmaps"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["endpoints"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["events"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["replicationcontrollers"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["replicationcontrollers/status"]
verbs: ["get"]
- apiGroups: [""]
resources: ["services"]
verbs: ["get", "list", "watch"]
- apiGroups: [""]
resources: ["services/status"]
verbs: ["get", "list", "watch"]
---
kind: RoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: rbac-role-binding
namespace: dev #指定 Namespace
subjects:
- kind: ServiceAccount
name: sa-dev #指定 ServiceAccount
namespace: dev #指定 Namespace
roleRef:
kind: Role
name: role-sa
apiGroup: rbac.authorization.k8s.io
(3)
授权namespace的权限,为什么要授权是因为sa内的secrets里的token只有在dashboard内使用,而上面的角色和角色绑定都是dev这个namespace内的,这样绑定后,拿到token才可以登录到dashboard的首页,否则都无法选择namespace。
[root@master sa]# cat cluster-role-binding.yaml
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: rbac-namespace-role
rules:
- apiGroups: [""] #配置权限,配置其只用于 namespace 的 list 权限
resources: ["namespaces"]
verbs: ["list"]
- apiGroups: [""]
resources: ["namespaces/status"]
verbs: ["get"]
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: rbac-default-role-binding
subjects:
- kind: ServiceAccount
name: sa-dev #配置为自定义的 ServiceAccount
namespace: dev #指定为服务账户所在的 Namespace
roleRef:
kind: ClusterRole
name: rbac-namespace-role #配置上面的 Role
apiGroup: rbac.authorization.k8s.io
(4)
单元测试
首先,使用deployment方式创建一个nginx的pod:
k create deploy nginx --image=nginx -n dev
获取这个sa下的secrets的token,使用该token登录dashboard:
[root@master sa]# kubectl -n dev describe secret $(kubectl get secret -n dev | grep sa-dev | awk '{print $1}')
Name: sa-dev-token-8ckbd
Namespace: dev
Labels: <none>
Annotations: kubernetes.io/service-account.name: sa-dev
kubernetes.io/service-account.uid: 7953d280-7b1a-4ba6-a0a4-e705e1cc9550
Type: kubernetes.io/service-account-token
Data
====
ca.crt: 1359 bytes
namespace: 3 bytes
token: eyJhbGciOiJSUzI1NiIsImtpZCI6IkxNRk93NFRwc3l5UlJjcG05V1IwY25rWjNGOWM1Z05OQjVXN1ROa004R1UifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJkZXYiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlY3JldC5uYW1lIjoic2EtZGV2LXRva2VuLThja2JkIiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZXJ2aWNlLWFjY291bnQubmFtZSI6InNhLWRldiIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50LnVpZCI6Ijc5NTNkMjgwLTdiMWEtNGJhNi1hMGE0LWU3MDVlMWNjOTU1MCIsInN1YiI6InN5c3RlbTpzZXJ2aWNlYWNjb3VudDpkZXY6c2EtZGV2In0.tGQMxNX-zbAAltH-GkgvDKhBm7VjvNDWwE77NLyD1naqsF8pMfd9_4MlSv9dHVe8KlRTaqH7AHVKvQnwuy68TKbFj-0Zzx7O5P7DI4Q7bmc2p_jwNxjX0RSARiTmk6pAaMN9tffH7FcwsmBKTDBKX7_X7e3MOrDeBsPLgqkFYAQk_bAld0Smv-HbYDuAw3WzdYsnOLmmW1ceUdZycPvHHmbccnhZUWFnjEx0lxdWBksmHJI60W1xAJNSv-EoKTz1klVaQgpCzJrXhv_MENPUeJgxPCS9o6nhoVG13s5gISf8aK7hHi9ccvtyWDsgFw7Od0Vd3x3IiK7o2IpPTormug
选择系统namespace kube-system以及其它任意的namespace,任何资源都看不到,除了dev这个namespace:
删除这个pod试试看(提示无权删除):
OK,编辑功能什么的都可以试一试,全部用不了,只有查看的权利,主要是因为前面的角色 role-sa 给的权限都是verbs: ["get", "watch", "list"]嘛,这样就完成了一个精细化的权限分配操作:指定的sa限定在一个指定的namespace里获取指定的资源(本例是指定的namespace里只有查看pod,deployment等个别资源的权限)。
小结:
那么,sa由于是给服务使用的,本例中是给dashboard使用的,这就造成了管理方面的局限性,总是要登录dashboard或者其它能够使用token的场景,sa的权限才会生效。那么,有没有给人使用的权限精细化分配方案呢?答案是必须有,通过kubeconfig文件就可以实现精准的权限管理啦,命令行和图形化界面都可以使用非常的方便哦。
三,
通过系统配置文件kubeconfig文件实现权限的精细化分配:
本次案例使用的集群为kubernetes二进制安装教程单master_zsk_john的博客-CSDN博客, 也就是集群的基本信息是:
token是前面的sa-dev这个sa里面的secrets,通过这个命令查询出来的:
kubectl -n dev describe secret $(kubectl get secret -n dev | grep sa-dev | awk '{print $1}')
kubeconfig文件生成前先把变量激活哦,这个不要忘记了:
KUBE_APISERVER="https://192.168.217.16:6443"
TOKEN="eyJhbGciOiJSUzI1NiIsImtpZCI6IkxNRk93NFRwc3l5UlJjcG05V1IwY25rWjNGOWM1Z05OQjVXN1ROa004R1UifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJkZXYiLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlY3JldC5uYW1lIjoic2EtZGV2LXRva2VuLThja2JkIiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZXJ2aWNlLWFjY291bnQubmFtZSI6InNhLWRldiIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50LnVpZCI6Ijc5NTNkMjgwLTdiMWEtNGJhNi1hMGE0LWU3MDVlMWNjOTU1MCIsInN1YiI6InN5c3RlbTpzZXJ2aWNlYWNjb3VudDpkZXY6c2EtZGV2In0.tGQMxNX-zbAAltH-GkgvDKhBm7VjvNDWwE77NLyD1naqsF8pMfd9_4MlSv9dHVe8KlRTaqH7AHVKvQnwuy68TKbFj-0Zzx7O5P7DI4Q7bmc2p_jwNxjX0RSARiTmk6pAaMN9tffH7FcwsmBKTDBKX7_X7e3MOrDeBsPLgqkFYAQk_bAld0Smv-HbYDuAw3WzdYsnOLmmW1ceUdZycPvHHmbccnhZUWFnjEx0lxdWBksmHJI60W1xAJNSv-EoKTz1klVaQgpCzJrXhv_MENPUeJgxPCS9o6nhoVG13s5gISf8aK7hHi9ccvtyWDsgFw7Od0Vd3x3IiK7o2IpPTormug"
(1)集群的ca证书生成
[root@master k8s-ssl]# cat ca-config.json
{
"signing": {
"default": {
"expiry": "87600h"
},
"profiles": {
"kubernetes": {
"expiry": "87600h",
"usages": [
"signing",
"key encipherment",
"server auth",
"client auth"
]
}
}
}
}
[root@master k8s-ssl]# cat ca-csr.json
{
"CN": "kubernetes",
"key": {
"algo": "rsa",
"size": 2048
},
"names": [
{
"C": "CN",
"L": "Beijing",
"ST": "Beijing","O": "k8s",
"OU": "System"
}
]
}
生成ca证书,证书生成后存放到任意目录下,例如/mnt:
cfssl gencert -initca ca-csr.json | cfssljson -bare ca -
cp ca.pem /mnt/
建立用户命令(这几个步骤是息息相关的,注意理解):
config文件引入集群ca证书,这里的set-cluster 可以任意设置,想叫什么集群名字都可以,我这里定义为mykubernetes,kubeconfig文件名称也随意定义,我这里定义为test.kubeconfig,此命令执行后会在当前目录生成test.kubeconfig这个文件:
kubectl config set-cluster mykubernetes \
--certificate-authority=/mnt/ca.pem \
--embed-certs=true \
--server= \
--kubeconfig=test.kubeconfig
定义的用户名称是test,当然,用户名称可以随意定义。里面的token就已经带了前面建立的sa的权限---查看dev这个namespace下的pod的权限:
kubectl config set-credentials "test" \
--token= \
--kubeconfig=test.kubeconfig
定义上下文的名称,这个也是随便定义,我就用了拼音权限分配,cluster 就是前面定义的集群名称 ,用户名称也是前面定义的,这个不能乱写哦,要匹配。
kubectl config set-context quanxianfenpei \
--cluster=mykubernetes \
--user=test \
--kubeconfig=test.kubeconfig
切换上下文,其实这一步就是将ca证书和token关联起来并写在了这个新定义的kubeconfig文件内。
kubectl config use-context quanxianfenpei \
--kubeconfig=test.kubeconfig
OK,这样我们就完成了使用kubeconfig配置精细化权限的流程,下面进入单元测试环节。
单元测试:
(1)
将test.kubeconfig这个文件拷贝到浏览器所在环境内,dashboard登录的时候选择此文件,成功登录dashboard,并且功能和上面的单元测试结果一致。
(2)
使用kubectl config 命令行测试:
可以查看pod,不能编辑删除pod
[root@master cfg]# k --kubeconfig=/root/kubeconfig/test.kubeconfig get po
[root@master cfg]# k --kubeconfig=/root/kubeconfig/test.kubeconfig delete pod
nginx-f89759699-cbnw6