1. DamonSet 方式部署 etcd

DaemonSet 的特性是在整个 Kubernetes 集群的所有节点上都启动一个副本，但在部署 etcd 时不能使用特性，这是需要结合 nodeSelector 将 DaemonSet 的 Pod 调度到固定的节点上，我们一般会选择固定的3台节点来调度 etcd 的 pod。

使用 DaemonSet 的好处是，不需要分别在3个节点上复制文件，部署相对简单。

在 Kubernetes 的 node 节点上配置 DaemonSet 调度匹配的 Label

kubectl label node 10.100.0.13 my-etcd=true

编辑 DaemonSet的 Yaml 文件

apiVersion: apps/v1
kind: DaemonSet
metadata:
  labels:
    app: my-etcd
  name: my-etcd
  namespace: kube-system
spec:
  selector:
    matchLabels:
      app: my-etcd
  template:
    metadata:
      labels:
        app: my-etcd
    spec:
      containers:
      - command:
        - /usr/local/bin/etcd
        - --advertise-client-urls=http://$(POD_IP):32379
        - --auto-compaction-mode=periodic
        - --auto-compaction-retention=1h
        - --client-cert-auth=false
        - --data-dir=/data/my-etcd-ds/data
        - --initial-advertise-peer-urls=http://$(POD_IP):32380
        - --initial-cluster-token=my-etcd
        - --listen-client-urls=http://$(POD_IP):32379,http://127.0.0.1:32379
        - --listen-peer-urls=http://$(POD_IP):32380
        - --logger=zap
        - --log-outputs=stderr,/data/my-etcd-ds/log/etcd.log
        - --name=$(NODE_NAME)
        - --peer-client-cert-auth=false
        - --snapshot-count=10000
        image: registry.cn-hangzhou.aliyuncs.com/cncfstack/mywiki-etcd:v1-20201229
        env:
        - name: ETCD_INITIAL_CLUSTER
          value: "10.100.0.13=http://10.100.0.13:32380"
        - name: ETCD_INITIAL_CLUSTER_STATE
          value: new
        - name: NODE_NAME
          valueFrom:
              fieldRef:
                fieldPath: spec.nodeName
        - name: POD_IP
          valueFrom:
            fieldRef:
              fieldPath: status.podIP
        livenessProbe:
          exec:
            command:
            - /bin/sh
            - -ec
            - ETCDCTL_API=3 etcdctl --endpoints=http://[127.0.0.1]:32379 get foo
          failureThreshold: 8
          initialDelaySeconds: 15
          timeoutSeconds: 15
        name: my-etcd
        resources:
          limits:
            cpu: "1"
            memory: "2G"
        volumeMounts:
        - mountPath: /data/my-etcd-ds/data
          name: etcd-data
        - mountPath: /etc/localtime
          name: localtime
        - mountPath: /data/my-etcd-ds/log
          name: etcd-log
        ports:
        - containerPort: 32379
          hostPort: 32379
        - containerPort: 32380
          hostPort: 32380
      hostNetwork: true
      nodeSelector:
        my-etcd: "true"
      volumes:
      - hostPath:
          path: /data/my-etcd-ds/data
        name: etcd-data
      - hostPath:
          path: /data/my-etcd-ds/log
        name: etcd-log
      - hostPath:
          path: /etc/localtime
        name: localtime
      dnsPolicy: ClusterFirstWithHostNet
      hostNetwork: true
      tolerations:
      - effect: NoSchedule
        operator: Exists
      - effect: NoExecute
        operator: Exists

这里主要关注集群主食和的配置 ETCD_INITIAL_CLUSTER ，名称设定和主机 IP 地址一致。

        - name: ETCD_INITIAL_CLUSTER
          value: "10.100.0.13=http://10.100.0.13:32380"
        - name: ETCD_INITIAL_CLUSTER_STATE
          value: new
        - name: NODE_NAME
          valueFrom:
              fieldRef:
                fieldPath: spec.nodeName
        - name: POD_IP
          valueFrom:
            fieldRef:
              fieldPath: status.podIP

创建 kubernetes 资源

$ kubectl  apply -f my-etcd-ds.yaml
daemonset.apps/my-etcd created

查看 pod 运行状态

$ kubectl  -n kube-system get pod |grep my-etcd
my-etcd-hdqh6                                     1/1     Running   0          3m24s

查看成员信息

$ etcdctl --endpoints http://127.0.0.1:32379 member list -w table
+------------------+---------+-------------+--------------------------+--------------------------+------------+
|        ID        | STATUS  |    NAME     |        PEER ADDRS        |       CLIENT ADDRS       | IS LEARNER |
+------------------+---------+-------------+--------------------------+--------------------------+------------+
| 3da87a463ee2e5a7 | started | 10.100.0.13 | http://10.100.0.13:32380 | http://10.100.0.13:32379 |      false |
+------------------+---------+-------------+--------------------------+--------------------------+------------+

读写数据

$ etcdctl --endpoints http://127.0.0.1:32379 put dstest dsvalue
OK

$ etcdctl --endpoints http://127.0.0.1:32379 get  dstest
dstest
dsvalue