从Kubernetes v1.3开始，DNS使用插件管理器集群附件自动启动的内置服务 。

正在运行的Kubernetes DNS pod 拥有3个容器：

  • kubedns：kubedns进程监视Kubernetes主服务器和端点的变化，并维护内存查找结构来服务于DNS请求。

  • dnsmasq：dnsmasq容器添加DNS缓存以提高性能。

  • healthz：healthz容器在执行双健康检查（ dnsmasq kubedns）时提供单个健康检查端点。

DNS pod作为具有静态IP的Kubernetes服务公开。一旦分配，kubelet将使用该--cluster-dns=<dns-service-ip> 标志配置的DNS传递给每个容器。

DNS名称也需要域名。本地域可以使用标志在kubelet中配置--cluster-domain=<default-local-domain>。

Kubernetes集群DNS服务器基于 SkyDNS库。它支持正向查找（A记录），服务查找（SRV记录）和反向IP地址查找（PTR记录）。

一、准备

1、镜像
andyshinn/dnsmasq:2.72
gcr.io/google_containers/cluster-proportional-autoscaler-amd64:1.1.1
gcr.io/google_containers/k8s-dns-kube-dns-amd64:1.14.7
gcr.io/google_containers/k8s-dns-sidecar-amd64:1.14.7
gcr.io/google_containers/k8s-dns-dnsmasq-nanny-amd64:1.14.7
2、创建文件夹
/etc/dnsmasq.d
/etc/dnsmasq.d-available
/etc/kubernetes/addons/dnsmasq/
/etc/kubernetes/addons/kube-dns/

二、配置文件

1、/etc/dnsmasq.d-available/01-kube-dns.conf
2、/etc/dnsmasq.d/01-kube-dns.conf          两个文件内容一样链接一下
#Listen on localhost
bind-interfaces
listen-address=0.0.0.0

addn-hosts=/etc/hosts

strict-order
# Forward k8s domain to kube-dns
server=/cluster.local/10.233.0.3
# Reply NXDOMAIN to bogus domains requests like com.cluster.local.cluster.local
local=/cluster.local.default.svc.cluster.local./default.svc.cluster.local.default.svc.cluster.local./com.default.svc.cluster.local./cluster.local.svc.cluster.local./svc.cluster.local.svc.cluster.local./com.svc.cluster.local./

#Set upstream dns servers
server=192.168.1.1

no-resolv
bogus-priv
no-negcache
cache-size=1000
dns-forward-max=150
max-cache-ttl=10
max-ttl=20
log-facility=-

三、部署dnsmasq

1、/etc/kubernetes/addons/dnsmasq/dnsmasq-clusterrolebinding.yml
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: dnsmasq
  namespace: "kube-system"
subjects:
  - kind: ServiceAccount
    name: dnsmasq
    namespace: "kube-system"
roleRef:
  kind: ClusterRole
  name: cluster-admin
  apiGroup: rbac.authorization.k8s.io
2、/etc/kubernetes/addons/dnsmasq/dnsmasq-serviceaccount.yml
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: dnsmasq
  namespace: "kube-system"
  labels:
    kubernetes.io/cluster-service: "true"
3、/etc/kubernetes/addons/dnsmasq/dnsmasq-deploy.yml
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: dnsmasq
  namespace: "kube-system"
  labels:
    k8s-app: dnsmasq
    kubernetes.io/cluster-service: "true"
spec:
  replicas: 1
  selector:
    matchLabels:
      k8s-app: dnsmasq
  strategy:
    type: "Recreate"
  template:
    metadata:
      labels:
        k8s-app: dnsmasq
        kubernetes.io/cluster-service: "true"
    spec:
      tolerations:
        - effect: NoSchedule
          operator: Exists
      containers:
        - name: dnsmasq
          image: "andyshinn/dnsmasq:2.72"
          imagePullPolicy: IfNotPresent
          command:
            - dnsmasq
          args:
            - -k
            - -C
            - /etc/dnsmasq.d/01-kube-dns.conf
          securityContext:
            capabilities:
              add:
                - NET_ADMIN
          resources:
            limits:
              cpu: 100m
              memory: 170Mi
            requests:
              cpu: 40m
              memory: 50Mi
          ports:
            - name: dns
              containerPort: 53
              protocol: UDP
            - name: dns-tcp
              containerPort: 53
              protocol: TCP
          volumeMounts:
            - name: etcdnsmasqd
              mountPath: /etc/dnsmasq.d
            - name: etcdnsmasqdavailable
              mountPath: /etc/dnsmasq.d-available
      volumes:
        - name: etcdnsmasqd
          hostPath:
            path: /etc/dnsmasq.d
        - name: etcdnsmasqdavailable
          hostPath:
            path: /etc/dnsmasq.d-available
      dnsPolicy: Default  # Don't use cluster DNS.
4、/etc/kubernetes/addons/dnsmasq/dnsmasq-svc.yml
---
apiVersion: v1
kind: Service
metadata:
  labels:
    kubernetes.io/cluster-service: 'true'
    k8s-app: dnsmasq
  name: dnsmasq
  namespace: kube-system
spec:
  ports:
    - port: 53
      name: dns-tcp
      targetPort: 53
      protocol: TCP
    - port: 53
      name: dns
      targetPort: 53
      protocol: UDP
  type: ClusterIP
  clusterIP: 10.233.0.2
  selector:
    k8s-app: dnsmasq
5、/etc/kubernetes/addons/dnsmasq/dnsmasq-autoscaler.yml
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: dnsmasq-autoscaler
  namespace: kube-system
  labels:
    k8s-app: dnsmasq-autoscaler
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
spec:
  template:
    metadata:
      labels:
        k8s-app: dnsmasq-autoscaler
      annotations:
        scheduler.alpha.kubernetes.io/critical-pod: ''
        scheduler.alpha.kubernetes.io/tolerations: '[{"key":"CriticalAddonsOnly", "operator":"Exists"}]'
    spec:
      serviceAccountName: dnsmasq
      tolerations:
        - effect: NoSchedule
          operator: Exists
      containers:
        - name: autoscaler
          image: gcr.io/google_containers/cluster-proportional-autoscaler-amd64:1.1.1
          resources:
            requests:
              cpu: "20m"
              memory: "10Mi"
          command:
            - /cluster-proportional-autoscaler
            - --namespace=kube-system
            - --configmap=dnsmasq-autoscaler
            - --target=Deployment/dnsmasq
            # When cluster is using large nodes(with more cores), "coresPerReplica" should dominate.
            # If using small nodes, "nodesPerReplica" should dominate.
            - --default-params={"linear":{"nodesPerReplica":10,"preventSinglePointFailure":true}}
            - --logtostderr=true
            - --v=2
6、创建
kubectl create -f /etc/kubernetes/addons/dnsmasq/

四、部署kube-dns

1、/etc/kubernetes/addons/kube-dns/kubedns-autoscaler-clusterrole.yml
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: cluster-proportional-autoscaler
  namespace: kube-system
rules:
  - apiGroups: [""]
    resources: ["nodes"]
    verbs: ["list"]
  - apiGroups: [""]
    resources: ["replicationcontrollers/scale"]
    verbs: ["get", "update"]
  - apiGroups: ["extensions"]
    resources: ["deployments/scale", "replicasets/scale"]
    verbs: ["get", "update"]
  - apiGroups: [""]
    resources: ["configmaps"]
    verbs: ["get", "create"]
2、/etc/kubernetes/addons/kube-dns/kubedns-autoscaler-clusterrolebinding.yml
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: cluster-proportional-autoscaler
  namespace: kube-system
subjects:
  - kind: ServiceAccount
    name: cluster-proportional-autoscaler
    namespace: kube-system
roleRef:
  kind: ClusterRole
  name: cluster-proportional-autoscaler
  apiGroup: rbac.authorization.k8s.io
3、/etc/kubernetes/addons/kube-dns/kubedns-autoscaler-sa.yml
---
kind: ServiceAccount
apiVersion: v1
metadata:
  name: cluster-proportional-autoscaler
  namespace: kube-system
4、/etc/kubernetes/addons/kube-dns/kubedns-autoscaler.yml
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: kubedns-autoscaler
  namespace: kube-system
  labels:
    k8s-app: kubedns-autoscaler
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
spec:
  template:
    metadata:
      labels:
        k8s-app: kubedns-autoscaler
    spec:
      tolerations:
        - effect: NoSchedule
          operator: Exists
      containers:
      - name: autoscaler
        image: "gcr.io/google_containers/cluster-proportional-autoscaler-amd64:1.1.1"
        resources:
          requests:
            cpu: "20m"
            memory: "10Mi"
        command:
        - /cluster-proportional-autoscaler
        - --namespace=kube-system
        - --configmap=kubedns-autoscaler
        # Should keep target in sync with cluster/addons/dns/kubedns-controller.yaml.base
        - --target=Deployment/kube-dns
        - --default-params={"linear":{"nodesPerReplica":10,"min":2}}
        - --logtostderr=true
        - --v=2
      serviceAccountName: cluster-proportional-autoscaler
5、/etc/kubernetes/addons/kube-dns/kubedns-deploy.yml
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: kube-dns
  namespace: "kube-system"
  labels:
    k8s-app: kube-dns
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
spec:
  # replicas: not specified here:
  # 1. In order to make Addon Manager do not reconcile this replicas parameter.
  # 2. Default is 1.
  # 3. Will be tuned in real time if DNS horizontal auto-scaling is turned on.
  strategy:
    rollingUpdate:
      maxSurge: 10%
      maxUnavailable: 0
  selector:
    matchLabels:
      k8s-app: kube-dns
  template:
    metadata:
      labels:
        k8s-app: kube-dns
      annotations:
        scheduler.alpha.kubernetes.io/critical-pod: ''
    spec:
      tolerations:
      - key: "CriticalAddonsOnly"
        operator: "Exists"
      - effect: NoSchedule
        operator: Exists
      volumes:
      - name: kube-dns-config
        configMap:
          name: kube-dns
          optional: true
      containers:
      - name: kubedns
        image: "gcr.io/google_containers/k8s-dns-kube-dns-amd64:1.14.7"
        imagePullPolicy: IfNotPresent
        resources:
          # TODO: Set memory limits when we've profiled the container for large
          # clusters, then set request = limit to keep this container in
          # guaranteed class. Currently, this container falls into the
          # "burstable" category so the kubelet doesn't backoff from restarting it.
          limits:
            memory: 170Mi
          requests:
            cpu: 40m
            memory: 50Mi
        livenessProbe:
          httpGet:
            path: /healthcheck/kubedns
            port: 10054
            scheme: HTTP
          initialDelaySeconds: 60
          timeoutSeconds: 5
          successThreshold: 1
          failureThreshold: 5
        readinessProbe:
          httpGet:
            path: /readiness
            port: 8081
            scheme: HTTP
          # we poll on pod startup for the Kubernetes master service and
          # only setup the /readiness HTTP server once that's available.
          initialDelaySeconds: 3
          timeoutSeconds: 5
        args:
        - --domain=cluster.local.
        - --dns-port=10053
        - --config-dir=/kube-dns-config
        - --v=2
        env:
        - name: PROMETHEUS_PORT
          value: "10055"
        ports:
        - containerPort: 10053
          name: dns-local
          protocol: UDP
        - containerPort: 10053
          name: dns-tcp-local
          protocol: TCP
        - containerPort: 10055
          name: metrics
          protocol: TCP
        volumeMounts:
        - name: kube-dns-config
          mountPath: /kube-dns-config
      - name: dnsmasq
        image: "gcr.io/google_containers/k8s-dns-dnsmasq-nanny-amd64:1.14.7"
        imagePullPolicy: IfNotPresent
        livenessProbe:
          httpGet:
            path: /healthcheck/dnsmasq
            port: 10054
            scheme: HTTP
          initialDelaySeconds: 60
          timeoutSeconds: 5
          successThreshold: 1
          failureThreshold: 5
        args:
        - -v=2
        - -logtostderr
        - -configDir=/etc/k8s/dns/dnsmasq-nanny
        - -restartDnsmasq=true
        - --
        - -k
        - --cache-size=1000
        - --log-facility=-
        - --server=/cluster.local/127.0.0.1#10053
        - --server=/in-addr.arpa/127.0.0.1#10053
        - --server=/ip6.arpa/127.0.0.1#10053
        ports:
        - containerPort: 53
          name: dns
          protocol: UDP
        - containerPort: 53
          name: dns-tcp
          protocol: TCP
        # see: https://github.com/kubernetes/kubernetes/issues/29055 for details
        resources:
          requests:
            cpu: 150m
            memory: 20Mi
        volumeMounts:
        - name: kube-dns-config
          mountPath: /etc/k8s/dns/dnsmasq-nanny
      - name: sidecar
        image: "gcr.io/google_containers/k8s-dns-sidecar-amd64:1.14.7"
        livenessProbe:
          httpGet:
            path: /metrics
            port: 10054
            scheme: HTTP
          initialDelaySeconds: 60
          timeoutSeconds: 5
          successThreshold: 1
          failureThreshold: 5
        args:
        - --v=2
        - --logtostderr
        - --probe=kubedns,127.0.0.1:10053,kubernetes.default.svc.cluster.local,5,A
        - --probe=dnsmasq,127.0.0.1:53,kubernetes.default.svc.cluster.local,5,A
        ports:
        - containerPort: 10054
          name: metrics
          protocol: TCP
        resources:
          requests:
            memory: 20Mi
            cpu: 10m
      dnsPolicy: Default  # Don't use cluster DNS.
      serviceAccountName: kube-dns
6、/etc/kubernetes/addons/kube-dns/kube-dns-sa.yml
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: kube-dns
  namespace: kube-system
  labels:
    kubernetes.io/cluster-service: "true"
---
apiVersion: v1
kind: Service
metadata:
  name: kube-dns
  namespace: kube-system
  labels:
    k8s-app: kube-dns
    kubernetes.io/cluster-service: "true"
    addonmanager.kubernetes.io/mode: Reconcile
    kubernetes.io/name: "KubeDNS"
spec:
  selector:
    k8s-app: kube-dns
  clusterIP: 10.233.0.3
  ports:
  - name: dns
    port: 53
    protocol: UDP
  - name: dns-tcp
    port: 53
    protocol: TCP

7、创建
kubectl create -f /etc/kubernetes/addons/kube-dns/

五、验证

2017-12-22 09-55-38 的屏幕截图.png