配置k8s dns

DNS (domain name system),提供域名解析服务，解决了难于记忆的IP地址问题，以更人性可读可记忆可标识的方式映射对应IP地址。

Cluster DNS扩展插件用于支持k8s集群系统中各服务之间发现与调用。

组件：
　　•SkyDNS 提供DNS解析服务
　　•Etcd 存储DNS信息
　　•Kube2sky 监听kubernetes，当有Service创建时，生成相应的记录到SkyDNS。

1.修改所有node的配置文件：/etc/kubernetes/kubelet kubernetes-node1，kubernetes-node2，kubernetes-node3:

[root@kubernetes-node2 ~]# tail -n 1 /etc/kubernetes/kubelet
KUBELET_ARGS="--cluster-dns=10.254.254.254 --cluster-domain=cluster.local"

2.所有node节点都重新启动服务

systemctl restart kubelet.service

3.根据namespace分为不同的名称空间，将系统服务统一放到一个单独的空间：kube-system，由于默认空间只有default，我们首先要创建kube-system空间。

[root@kubernetes-master pods]# cat kube-system.yml
apiVersion: v1
kind: Namespace
metadata:
name: kube-system

4.在master服务器上编辑ReplicationController文件dns-rc.yml

  1 apiVersion: v1
  2 kind: ReplicationController
  3 metadata:
  4   name: kube-dns-v11
  5   namespace: kube-system
  6   labels:
  7     k8s-app: kube-dns
  8     version: v11
  9     kubernetes.io/cluster-service: "true"
 10 spec:
 11   replicas: 1
 12   selector:
 13     k8s-app: kube-dns
 14     version: v11
 15   template:
 16     metadata:
 17       labels:
 18         k8s-app: kube-dns
 19         version: v11
 20         kubernetes.io/cluster-service: "true"
 21     spec:
 22       containers:
 23       - name: etcd
 24         image: docker.gaoxiaobang.com/kubernetes/etcd-amd64:2.2.1  
 25         resources:
 26           limits:
 27             cpu: 100m
 28             memory: 500Mi
 29           requests:
 30             cpu: 100m
 31             memory: 50Mi
 32         command:
 33         - /usr/local/bin/etcd
 34         - -data-dir
 35         - /var/etcd/data
 36         - -listen-client-urls
 37         - http://127.0.0.1:2379,http://127.0.0.1:4001
 38         - -advertise-client-urls
 39         - http://127.0.0.1:2379,http://127.0.0.1:4001
 40         - -initial-cluster-token
 41         - skydns-etcd
 42         #volumeMounts:
 43         #- name: etcd-storage
 44         #  mountPath: /var/etcd/data
 45       - name: kube2sky
 46         image: docker.gaoxiaobang.com/kubernetes/kube2sky:1.14
 47         args:
 48         # command = "/kube2sky"
 49         - --domain=cluster.local
 50         - --kube-master-url=http://10.64.8.68:8080
 51         resources:
 52           limits:
 53             cpu: 100m
 54             memory: 200Mi
 55           requests:
 56             cpu: 100m
 57             memory: 50Mi
 58       - name: skydns
 59         image: docker.gaoxiaobang.com/kubernetes/skydns:2015-10-13-8c72f8c
 60         resources:
 61           limits:
 62             cpu: 100m
 63             memory: 200Mi
 64           requests:
 65             cpu: 100m
 66             memory: 50Mi
 67         args:
 68         - -machines=http://127.0.0.1:4001
 69         - -addr=0.0.0.0:53
 70         - -ns-rotate=false
 71         - -domain=cluster.local.
 72         ports:
 73         - containerPort: 53
 74           name: dns
 75           protocol: UDP
 76         - containerPort: 53
 77           name: dns-tcp
 78           protocol: TCP
 79         livenessProbe:
 80           httpGet:
 81             path: /healthz
 82             port: 8080
 83             scheme: HTTP
 84           initialDelaySeconds: 60
 85           timeoutSeconds: 5
 86           successThreshold: 1
 87           failureThreshold: 5
 88         readinessProbe:
 89           httpGet:
 90             path: /readiness
 91             port: 8080
 92             scheme: HTTP
 93           initialDelaySeconds: 30
 94           timeoutSeconds: 5
 95       - name: healthz
 96         image: docker.gaoxiaobang.com/kubernetes/exechealthz:1.0
 97         resources:
 98           limits:
 99             cpu: 10m
100             memory: 20Mi
101           requests:
102             cpu: 10m
103             memory: 20Mi
104         args:
105         - -cmd=nslookup kubernetes.default.svc.cluster.local localhost >/dev/null
106         - -port=8080
107         ports:
108         - containerPort: 8080
109           protocol: TCP
110       volumes:
111       - name: etcd-storage
112         emptyDir: {}
113       dnsPolicy: Default

dns-rc.yml

注意dns-rc.yml：修改文件中的master ip

dns service服务yml

 1 apiVersion: v1
 2 kind: Service
 3 metadata:
 4   name: kube-dns
 5   namespace: kube-system
 6   labels:
 7     k8s-app: kube-dns
 8     kubernetes.io/cluster-service: "true"
 9     kubernetes.io/name: "KubeDNS"
10 spec:
11   selector:
12     k8s-app: kube-dns
13   clusterIP:  10.254.254.254
14   ports:
15   - name: dns
16     port: 53
17     protocol: UDP
18   - name: dns-tcp
19     port: 53
20     protocol: TCP

dns-svc.yml

创建namespace（有就不用在创建了）

kubectl create -f kube-system.yml

创建ReplicationController

[root@kubernetes-master pods]# kubectl create -f dns-rc.yml

创建Service

[root@kubernetes-master pods]# kubectl create -f dns-svc.yml

查看dns-rc.yml文件中定义pod的启动状态，如下则启动成功。

[root@kubernetes-master pods]# kubectl get pod --namespace=kube-system

查看dns-svc.yml文件中定义的Service的信息。

[root@kubernetes-master pods]# kubectl get svc -l k8s-app=kube-dns --namespace=kube-system

测试DNS的效果。
部署一个busybox的Pod

 1 apiVersion: v1
 2 kind: Pod
 3 metadata:
 4   name: busybox
 5   namespace: default
 6 spec:
 7   containers:
 8   - image: busybox
 9     command:
10       - sleep
11       - "3600"
12     imagePullPolicy: IfNotPresent
13     name: busybox
14   restartPolicy: Always

busybox.yml

验证kubernetes服务的地址解析，出现如下所示则证明DNS解析正常。