背景
cks考试资格是去年活动时候跟cka一起买的 1200左右大洋吧…考了两次 ,第一次57分。我考!第二次 62分, 竟然还是没有过去…可能冥冥之中自己有所感应,今年活动的时候购买了一次机会备用的…好歹第三次算是过了86分还好…
总结一下15个题吧!
具体可参考昕光xg大佬的博客CKS认证–CKS 2021最新真题考试经验分享–练习题04。顺序记不太清了,就按照昕光xg大佬列的题目记录一下解题思路吧!墙裂推荐大佬的博客。满满的都是干货!
1. RuntimeClass gVisor
根据题目内容创建一个RuntimeClass,然后修改统一namespace下的pod
参照:官方文档https://kubernetes.io/docs/concepts/containers/runtime-class/#usage%3Cbr%3E
vim /home/cloud_user/sandbox.yml
apiVersion: node.k8s.io/v1
kind: RuntimeClass
metadata:
name: sandbox
handler: runsc
kubectl apply -f /home/xxx/sandbox.yml
修改xxxx namespace下 pod的runtimeClassName: sandbox
kubectl -n xxx edit deployments.apps work01 # runtimeClassName: sandbox
kubectl -n xxx edit deployments.apps work02
kubectl -n xxx edit deployments.apps work03
基本就是这样的 修改3个deployments。增加runtimeClassName: sandbox!等待pod重建完成!
如何确定修改成功呢?
kubectl -n xxx exec xxx – dmesg
or kubectl get deployments xxx -n xxx -o yaml|grep runtime就可以吧?
2. NetworkPolicy, 限制指定pod、ns 访问指定labels的一组pod
这个题貌似一直是没有变的,网上也看了好多的解题方法 但是貌似都是有问题的
在development命名空间内,创建名为pod-access的NetworkPolicy作用于名为products-service的Pod,只允许命名空间为test的Pod或者在任何命名空间内有environment:staging标签的Pod来访问。
当然了namespace 和podname networkpolicy的名字会变
重要的是确定pod namespace的labels
kubectl get pods -n development --show-labels
kubectl get ns --show-labels
cat network-policy.yaml
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: pod-access
namespace: development
spec:
podSelector:
matchLabels:
environment: staging
policyTypes:
- Ingress
ingress:
- from: #命名空间有name: testing标签的Pod
- namespaceSelector:
matchLabels:
name: testing
- from: #所有命名空间有environment: staging标签的Pod
- namespaceSelector:
matchLabels:
podSelector:
matchLabels:
environment: staging
kubectl apply -f network-policy.yaml
官方文档参考:https://kubernetes.io/docs/concepts/services-networking/network-policies/
3. NetworkPolicy,dany all Ingress Egress
考了两次,这个题会变 但是无非是创建名为 denynetwork 的 NetworkPolicy,拒绝development命名空间内所有Ingress流量or Egress or Ingress and Egress流量
参照:https://kubernetes.io/docs/concepts/services-networking/network-policies/#default-deny-all-ingress-and-all-egress-traffic
注意审题看题目中的要求!
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: denynetwork
namespace: development
spec:
podSelector: {}
policyTypes:
- Ingress
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: denynetwork
namespace: development
spec:
podSelector: {}
policyTypes:
- Egress
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: denynetwork
namespace: development
spec:
podSelector: {}
policyTypes:
- Ingress
- Egress
4. trivy 检测镜像
kubectl -n development get pods
kubectl -n development get pods --output=custom-columns="NAME:.metadata.name,IMAGE:.spec.containers[*].image"
NAME IMAGE
work1 busybox:1.33.1
work2 nginx:1.14.2
work3 amazonlinux:2
work4 amazonlinux:1
work5 centos:7
trivy image -s HIGH,CRITICAL busybox:1.33.1
trivy image -s HIGH,CRITICAL nginx:1.14.2 #HIGH and CRITICAL
trivy image -s HIGH,CRITICAL amazonlinux:2
trivy image -s HIGH,CRITICAL amazonlinux:1
trivy image -s HIGH,CRITICAL centos:7 #HIGH and CRITICAL
我考试题目中是有两个镜像 centos的镜像不符合要求把对应pod删除即可
https://kubernetes.io/zh/docs/reference/kubectl/cheatsheet/#%E6%A0%BC%E5%BC%8F%E5%8C%96%E8%BE%93%E5%87%BA
5. kube-bench,修复不安全项
这个题目也比较稳就是根据bue-bench对master节点进行修复,基本就是老一套 。根据题目中提示基本没有什么问题!
kube-apiserver
vim /etc/kubernetes/manifests/kube-apiserver.yaml
- --authorization-mode=Node,RBAC
kubelet
vim /var/lib/kubelet/config.yaml
authentication:
anonymous:
enabled: false
webhook:
enabled: true
authorization:
mode: Webhook
protectKernelDefaults: true
systemctl restart kubelet.service
systemctl status kubelet.service
etcd
mv /etc/kubernetes/manifests/etcd.yaml /etc/kubernetes/
vim /etc/kubernetes/etcd.yaml
- --client-cert-auth=true
这个题也很刺激 。我进入master节点发现kuberentes环境竟然没有启动…what.我修改了配置文件后还是没有起来…中间我的vpn断网了。然后检查的时候环境竟然好了。莫名其妙的…
6. clusterrole
这里考了两次也都会细微的改变。但是无非是以下三个步骤
- 修改namespace下role的权限只允许对某一类对象做list list的操作
- 新建一个serviceaccount
- 创建名为role-2的role,并且通过rolebinding绑定sa-dev-1,只允许对persistentvolumeclaims做update操作。
# kubectl -n db get role
NAME ROLE AGE
sa-dev-1 Role/role-1 7d16h
# kubectl -n db edit role role-1
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
name: role-1
namespace: db
selfLink: /apis/rbac.authorization.k8s.io/v1/namespaces/db/roles/role-1
rules:
- apiGroups:
- ""
resources:
- endpoints #只允许对endpoints资源list
verbs:
- list
kubectl -n db create role role-2 --resource=persistentvolumeclaims --verb=update
kubectl create rolebinding role-2-binding --role=role-2 --serviceaccount=db:service-account-web -n db
7. serviceAccount
在qa命名空间内创建ServiceAccount frontend-sa,不允许访问任何secrets。创建名为frontend-sa的Pod使用该ServiceAccount。并删除qa命名空间下没有使用的sa
参照:https://kubernetes.io/docs/tasks/configure-pod-container/configure-service-account/
apiVersion: v1
kind: ServiceAccount
metadata:
name: frontend-sa
namespace: qa
automountServiceAccountToken: false
...
apiVersion: v1
kind: Pod
metadata:
name: my-pod
namespace: qa
spec:
serviceAccountName: frontend-sa
automountServiceAccountToken: false
...
kubectl get sa -n qa 应该是有三个sa frontend-sa default 还有另外一个 正常应该是保留frontend-sa 删除另外两个?但是default有必要删除吗?哈哈哈哈这个地方有点徘徊!我貌似
8. Dockerfile 和 Pod yaml 检测
这个地方也有点郁闷啊。不是每个文件中让修改两处吗?Dockerfile中我是不是要修改三处呢?题目中注明了 基础镜像是ubuntu:16.04 但是给的文件中是latest应该是要修改的吧?
Dockerfile:
去掉两处 USER root
设置基础镜像为 ubuntu:16.04
Pod yaml:
注释掉 privileged 那一行的相关配置
https://docs.docker.com/develop/develop-images/dockerfile_best-practices/#from
9. secret
istio-system 命名空间中有一个名称为 db1-test 的 secret, 按照要求完成如下内容:
- 存储 username 字段到 /home/candidate/user.txt 文件 , password 字段到 /home/candidate/old_pass.txt 文件。用户需要自己创建文件。
- 创建secret from user password。
- 将secret挂载到pod中。
在 istio-system 命名空间创建名称为 db2-test 的 secret,其包含如下内容:
pod name secret-pod
namespace istio-system
container name dev-container
image nginx ###也可能是httpd我考的几次都碰到过
volume name secret-volume
mount path /etc/secret
1. 将secrets base 64写入文件
很多解题都用了kubectl jsonpath的方式个人比较笨就用了传统的方式:
kubectl get secrets -n istio-system db1-test -o yaml
echo "" |base64 -d > /home/candidate/old_pass.txt
echo "" |base64 -d > /home/candidate/user.txt
注意顺序 查看secrets的时候貌似password都在上面,个人习惯总是容易先搞username。还是容易搞混的
2. 生成secret
参照:https://kubernetes.io/docs/tasks/configmap-secret/managing-secret-using-kubectl/
kubectl create secret generic db2-test -n istio-system \
--from-literal=username=production-instance \
--from-literal=password=KvLftKgs4aHV
3 . secret 挂载到pod中
apiVersion: v1
kind: Pod
metadata:
name: secret-pod
namespace: istio-system
spec:
containers:
- name: dev-container
image: nginx ###我的是httpd
volumeMounts:
- name: secret-volume
mountPath: "/etc/secret"
readOnly: true
volumes:
- name: secret-volume
secret:
secretName: db2-test
10. PodSecurityPolicy
基本就是参照官方:
- 创建名为 restrict-policy的PodSecurityPolicy,阻止创建privileged Pod
- 创建名为 restrict-access-role 的ClusterRole允许使用新创建的名为 restrict-policy 的PodSecurityPolicy。
- 在staging命名空间中创建名为 psp-denial-sa 的serviceAccount。
- 创建名为deny-access-bind的clusterRoleBinding,绑定刚刚创建的serviceAccount和ClusterRole。
vim restrict-policy.yml
apiVersion: policy/v1beta1
kind: PodSecurityPolicy
metadata:
name: restrict-policy
spec:
privileged: false ###一定要记得这里false吧?
runAsUser:
rule: "RunAsAny"
fsGroup:
rule: "RunAsAny"
seLinux:
rule: "RunAsAny"
supplementalGroups:
rule: "RunAsAny"
kubectl apply -f restrict-policy.yml
kubectl create clusterrole restrict-access-role --verb=use --resource=psp --resource-name=restrict-policy
kubectl create sa psp-denial-sa -n staging
kubectl create clusterrolebinding deny-access-bind --clusterrole=restrict-access-role --serviceaccount=staging:psp-denial-sa
注:参考https://kubernetes.io/docs/concepts/policy/pod-security-policy/#volumes-and-file-systems
11. Container(容器)安全 ,删除包含Volumes或privileged 的pod
查看production namespace下所有pod是否有特权 Privileged 或者挂载 volume 的 pod
kubectl get pods NAME -n production -o jsonpath={.spec.volumes} | jq
kubectl get pods NAME -o yaml -n production | grep "privi.*: true"
然后删除 Privileged 或者挂载 volume 的 Pod。应该是有3个pod 要删除两个?
12. audit–日志审计
日志审计这个地方很刺激。不知道为什么我改了前两次都没有起来…
现在回想一下我默认的人家写好的都删了这应该是不对的:
** basic policy is provided at /etc/kubernetes/logpolicy’sample-policy.yaml . It onlyspecifies what not to log----指定不记录到日志中的内容**
- namespaces changes at RequestResponse level
- the request body of persistentwolumes changes inthe namespace front-apps
- ConfigMap and Secret changes in all namespaces atthe Metadata level
cat /etc/kubernetes/logpolicy/sample-policy.yaml
apiVersion: audit.k8s.io/v1 # This is required.
kind: Policy
omitStages:
- "RequestReceived"
rules:
# 保留指定不记录到日志中的内容(原配置文件内容)
- level: RequestResponse
resources:
- group: ""
resources: ["namespaces"]
- level: Request
resources:
- group: ""
resources: ["persistentvolumes"]
namespaces: ["front-apps"]
- level: Metadata
resources:
- group: "" # core API group
resources: ["secrets", "configmaps"]
- level: Metadata
omitStages:
- "RequestReceived"
kube-apiserver.yaml
配置文件路径,log路径 最大保留天数,保留的审计日志文件的最大数量
vim /etc/kubernetes/manifests/kube-apiserver.yaml
- --audit-policy-file=/etc/kubernetes/audit-policy.yaml
- --audit-log-path=/var/log/kubernetes/audit.log
- --audit-log-maxage=10
- --audit-log-maxbackup=1
重启服务并验证
systemctl restart kubelet
kubectl apply -f xxx.yaml
tail -f /var/log/kubernetes/audit.log
注: 修改之前先做好备份!谨记参照官方:https://kubernetes.io/docs/tasks/debug-application-cluster/audit/。文件的挂载环境中已配置好了不用修改添加的。
13. image policy
参照:https://kubernetes.io/docs/reference/access-authn-authz/admission-controllers/
admission-control.conf or|dmission-control.yaml
vim /etc/kubernetes/admission-control/admission-control.conf
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: ImagePolicyWebhook
path: imagepolicy.conf
imagepolicy.conf | imagepolicy.json
vim /etc/kubernetes/admission-control/imagepolicy.conf
{
"imagePolicy": {
"kubeConfigFile": "/etc/kubernetes/admission-control/imagepolicy_backend.kubeconfig",
"allowTTL": 50,
"denyTTL": 50,
"retryBackoff": 500,
"defaultAllow": false
}
}
注: 只修改了defaultAllow true为false!
imagepolicy_backend.kubeconfig
vim /etc/kubernetes/admission-control/imagepolicy_backend.kubeconfig
apiVersion: v1
kind: Config
clusters:
- name: trivy-k8s-webhook
cluster:
certificate-authority: /etc/kubernetes/admission-control/imagepolicywebhook-ca.crt
server: https://acg.trivy.k8s.webhook:8090/scan
contexts:
- name: trivy-k8s-webhook
context:
cluster: trivy-k8s-webhook
user: api-server
current-context: trivy-k8s-webhook
preferences: {}
users:
- name: api-server
user:
client-certificate: /etc/kubernetes/admission-control/api-server-client.crt
client-key: /etc/kubernetes/admission-control/api-server-client.key
#只在server中增加了相关配置!
kube-apiserver
vim /etc/kubernetes/manifests/kube-apiserver.yaml
- --admission-control-config-file=/etc/kubernetes/admission-control/admission-control.conf
- --enable-admission-plugins=NodeRestriction,ImagePolicyWebhook
验证
systemctl restart kubelet
kubectl apply -f /root/xxx/vulnerable-manifest.yaml
tail -n 10 /var/log/imagepolicy/roadrunner.log
14. api-server 参数调整
这个题没有读懂。没有搞明白 只看懂了apiserver中修改了:
- --enable-admission-plugins=AlwaysAdmit
修改为:
- --enable-admission-plugins=NodeRestriction
然后根据题目提示还删除了了一个匿名的clusterrole…接下来就不知道要怎么搞了。正常就是这个题得分应该是没有的
15. AppArmor
注: 参照https://kubernetes.io/docs/tutorials/clusters/apparmor/
ssh到work节点
cat /etc/apparmor.d/nginx_apparmor
#include <tunables/global>
profile nginx-profile-3 flags=(attach_disconnected) {
#include <abstractions/base>
file,
# Deny all file writes.
deny /** w,
}
sudo apparmor_status | grep nginx
sudo -qa /etc/apparmor.d/nginx_apparmor
sudo apparmor_status | grep nginx
nginx-profile-3
exit work节点回到跳板机:
vim https://github.com/lmtbelmonte/cks-cert
apiVersion: v1
kind: Pod
metadata:
name: writedeny
namespace: dev
annotations:
container.apparmor.security.beta.kubernetes.io/busybox: localhost/nginx-profile-3
spec:
containers:
- name: busybox
image: busybox:1.33.1
kubecl apply -f ~/nginx-deploy.yml
其他:
相关资料参考:
- 昕光xg大佬的博客https://blog.csdn.net/u011127242/category_10823035.html
- https://github.com/ggnanasekaran77/cks-exam-tips
- https://github.com/jayendrapatil/kubernetes-exercises/tree/main/topics
- https://github.com/lmtbelmonte/cks-cert
- https://github.com/PatrickPan93/cks-relative
- https://github.com/moabukar/CKS-Exercises-Certified-Kubernetes-Security-Specialist
以上几个资源刷一遍 cks必过!。当然了还有官方文档:https://kubernetes.io/docs/concepts/