Kubeadm HA 1.9 高可用 集群 本地离线部署

k8s介绍

• k8s 发展速度很快，目前很多大的公司容器集群都基于该项目，如京东，腾讯，滴滴，瓜子二手车，易宝支付，北森等等。
  
• kubernetes1.9版本发布2017年12月15日，每三个月一个迭代版本， Workloads API成为稳定版本，这消除了很多潜在用户对于该功能稳定性的担忧。还有一个重大更新，就是测试支持了Windows了，这打开了在kubernetes中运行Windows工作负载的大门。
• CoreDNS alpha可以使用标准工具来安装CoreDNS
• kube-proxy的IPVS模式进入beta版，为大型集群提供更好的可扩展性和性能。
• kube-router的网络插件支持，更方便进行路由控制，发布，和安全策略管理

---

k8s 核心架构

如架构图

• k8s 高可用2个核心 ==apiserver master== and ==etcd==
• ==apiserver master==：（需高可用）集群核心，集群API接口、集群各个组件通信的中枢；集群安全控制；
• ==etcd== ：（需高可用）集群的数据中心，用于存放集群的配置以及状态信息，非常重要，如果数据丢失那么集群将无法恢复；因此高可用集群部署首先就是etcd是高可用集群；
• kube-scheduler：调度器 （内部自选举）集群Pod的调度中心；默认kubeadm安装情况下--leader-elect参数已经设置为true，保证master集群中只有一个kube-scheduler处于活跃状态；
• kube-controller-manager： 控制器 （内部自选举）集群状态管理器，当集群状态与期望不同时，kcm会努力让集群恢复期望状态，比如：当一个pod死掉，kcm会努力新建一个pod来恢复对应replicas set期望的状态；默认kubeadm安装情况下--leader-elect参数已经设置为true，保证master集群中只有一个kube-controller-manager处于活跃状态；
• kubelet: agent node注册apiserver
• kube-proxy: 每个node上一个，负责service vip到endpoint pod的流量转发，老版本主要通过设置iptables规则实现，新版1.9基于kube-proxy-lvs 实现 更好性能和负载拓展

---

部署示意图

• 集群ha方案，我们力求简单，使用keepalive 监听一个vip来实现，（当节点不可以后，会有vip漂移的切换时长，取决于我们设置timeout切换时长，测试会有10s空档期，如果对高可用更高要求 可以用lvs或者nginx做 4层lb负载 更佳完美，我们力求简单够用，可接受10s的api不可用）
• etcd和master 测试采用 3节点共用部署，（生成环境可以考虑etcd和master分离或者等kubeadm ha正式版推出）

---

部署环境

最近在部署k8s 1.9集群遇到一些问题，整理记录，或许有助需要的朋友。
因为kubeadm 简单便捷，所以集群基于该项目部署，目前bete版本不支持ha部署，github说2018年预计发布ha 版本，可我们等不及了 呼之欲来。。。

---

环境	版本
Centos	CentOS Linux release 7.3.1611 (Core)
Kernel	Linux etcd-host1 3.10.0-514.el7.x86_64
yum base repo	http://mirrors.aliyun.com/repo/Centos-7.repo
yum epel repo	http://mirrors.aliyun.com/repo/epel-7.repo
kubectl	v1.9.0
kubeadmin	v1.9.0
docker	1.12.6
docker localre	devhub.beisencorp.com

---

主机名称	相关信息	备注
etcd-host1	10.129.6.211	master和etcd
etcd-host2	10.129.6.212	master和etcd
etcd-host3	10.129.6.213	master和etcd
Vip-keepalive	10.129.6.220	vip用于高可用

环境部署 （我们使用本地离线镜像）

环境预初始化

• Centos Mini安装 每台机器root
• 设置机器名

hostnamectl set-hostname etcd-host1

• 停防火墙

systemctl stop firewalld
systemctl disable firewalld
systemctl disable firewalld

• 关闭Swap

swapoff -a
sed 's/.*swap.*/#&/' /etc/fstab

• 关闭防火墙

systemctl disable firewalld && systemctl stop firewalld && systemctl status firewalld

• 关闭Selinux

setenforce  0
sed -i "s/^SELINUX=enforcing/SELINUX=disabled/g" /etc/sysconfig/selinux
sed -i "s/^SELINUX=enforcing/SELINUX=disabled/g" /etc/selinux/config
sed -i "s/^SELINUX=permissive/SELINUX=disabled/g" /etc/sysconfig/selinux
sed -i "s/^SELINUX=permissive/SELINUX=disabled/g" /etc/selinux/config getenforce

• 增加DNS

echo nameserver 114.114.114.114>>/etc/resolv.conf

• 设置内核

cat <<EOF >  /etc/sysctl.d/k8s.conf
net.bridge.bridge-nf-call-ip6tables = 1
net.bridge.bridge-nf-call-iptables = 1
EOF
sysctl -p /etc/sysctl.conf#若问题
执行sysctl -p 时出现：
sysctl -p
sysctl: cannot stat /proc/sys/net/bridge/bridge-nf-call-ip6tables: No such file or directory
sysctl: cannot stat /proc/sys/net/bridge/bridge-nf-call-iptables: No such file or directory解决方法：
modprobe br_netfilter
ls /proc/sys/net/bridge

配置keepalived

• VIP Master 通过控制VIP 来HA高可用（常规方案）
• 到目前为止,三个master节点 相互独立运行,互补干扰. kube-apiserver作为核心入口, 可以使用keepalived 实现高可用, kubeadm join暂时不支持负载均衡的方式,所以我们
• 安装

yum install -y keepalived

• 配置keepalived.conf

cat >/etc/keepalived/keepalived.conf  <<EOL
global_defs {router_id LVS_k8s
}vrrp_script CheckK8sMaster {script "curl -k https://10.129.6.220:6443"interval 3timeout 9fall 2rise 2
}vrrp_instance VI_1 {state MASTERinterface ens32virtual_router_id 61# 主节点权重最高 依次减少priority 120advert_int 1#修改为本地IP mcast_src_ip 10.129.6.211nopreemptauthentication {auth_type PASSauth_pass sqP05dQgMSlzrxHj}unicast_peer {#注释掉本地IP#10.129.6.21110.129.6.21210.129.6.213}virtual_ipaddress {10.129.6.220/24}track_script {CheckK8sMaster}}
EOL

• 启动

systemctl enable keepalived && systemctl restart keepalived

• 结果

[root@etcd-host1 k8s]# systemctl status keepalived
● keepalived.service - LVS and VRRP High AvailabilitymonitorLoaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)Active: active (running) since Fri 2018-01-19 10:27:58 CST; 8h agoMain PID: 1158 (keepalived)CGroup: /system.slice/keepalived.service├─1158 /usr/sbin/keepalived -D├─1159 /usr/sbin/keepalived -D└─1161 /usr/sbin/keepalived -DJan 19 10:28:00 etcd-host1 Keepalived_vrrp[1161]: Sending gratuitous ARP on ens32 for 10.129.6.220
Jan 19 10:28:05 etcd-host1 Keepalived_vrrp[1161]: VRRP_Instance(VI_1) Sending/queueing gratuitous ARPs on ens32 for 10.129.6.220

• 依次配置 其他2台从节点master 配置 修改对应节点 ip
• master01 priority 120
• master02 priority 110
• master03 priority 100

Etcd https 集群部署

Etcd 环境准备

#机器名称
etcd-host1：10.129.6.211
etcd-host2：10.129.6.212
etcd-host3：10.129.6.213#部署环境变量
export NODE_NAME=etcd-host3 #当前部署的机器名称(随便定义，只要能区分不同机器即可)
export NODE_IP=10.129.6.213 # 当前部署的机器 IP
export NODE_IPS="10.129.6.211 10.129.6.212 10.129.6.213" # etcd 集群所有机器 IP
# etcd 集群间通信的IP和端口
export ETCD_NODES=etcd-host1=https://10.129.6.211:2380,etcd-host2=https://10.129.6.212:2380,etcd-host3=https://10.129.6.213:2380

Etcd 证书创建（我们使用https方式）

创建 CA 证书和秘钥

• 安装cfssl， CloudFlare 的 PKI 工具集 cfssl 来生成 Certificate Authority (CA) 证书和秘钥文件
• 如果不希望将cfssl工具安装到部署主机上，可以在其他的主机上进行该步骤，生成以后将证书拷贝到部署etcd的主机上即可。本教程就是采取这种方法，在一台测试机上执行下面操作。

wget https://pkg.cfssl.org/R1.2/cfssl_linux-amd64
chmod +x cfssl_linux-amd64
mv cfssl_linux-amd64 /usr/local/bin/cfsslwget https://pkg.cfssl.org/R1.2/cfssljson_linux-amd64
chmod +x cfssljson_linux-amd64
mv cfssljson_linux-amd64 /usr/local/bin/cfssljsonwget https://pkg.cfssl.org/R1.2/cfssl-certinfo_linux-amd64
chmod +x cfssl-certinfo_linux-amd64
mv cfssl-certinfo_linux-amd64 /usr/local/bin/cfssl-certinfo

生成ETCD的TLS 秘钥和证书

• 为了保证通信安全，客户端(如 etcdctl) 与 etcd 集群、etcd 集群之间的通信需要使用 TLS 加密，本节创建 etcd TLS 加密所需的证书和私钥。

• 创建 CA 配置文件：

  cat >  ca-config.json <<EOF
  {
  "signing": {
  "default": {"expiry": "8760h"
  },
  "profiles": {"kubernetes": {"usages": ["signing","key encipherment","server auth","client auth"],"expiry": "8760h"}
  }
  }
  }
  EOF

• ==ca-config.json==：可以定义多个 profiles，分别指定不同的过期时间、使用场景等参数；后续在签名证书时使用某个 profile；
• ==signing==：表示该证书可用于签名其它证书；生成的 ca.pem 证书中 CA=TRUE；
• ==server auth==：表示 client 可以用该 CA 对 server 提供的证书进行验证；
• ==client auth==：表示 server 可以用该 CA 对 client 提供的证书进行验证；

  cat >  ca-csr.json <<EOF
  {
  "CN": "kubernetes",
  "key": {
  "algo": "rsa",
  "size": 2048
  },
  "names": [
  {"C": "CN","ST": "BeiJing","L": "BeiJing","O": "k8s","OU": "System"
  }
  ]
  }
  EOF

• "CN"：Common Name，kube-apiserver 从证书中提取该字段作为请求的用户名 (User Name)；浏览器使用该字段验证网站是否合法；
• "O"：Organization，kube-apiserver 从证书中提取该字段作为请求用户所属的组 (Group)；

• ==生成 CA 证书和私钥==：

cfssl gencert -initca ca-csr.json | cfssljson -bare ca
ls ca*

==创建 etcd 证书签名请求：==

cat > etcd-csr.json <<EOF
{"CN": "etcd","hosts": ["127.0.0.1","10.129.6.211","10.129.6.212","10.129.6.213"],"key": {"algo": "rsa","size": 2048},"names": [{"C": "CN","ST": "BeiJing","L": "BeiJing","O": "k8s","OU": "System"}]
}
EOF

• hosts 字段指定授权使用该证书的 etcd 节点 IP；
• 每个节点IP 都要在里面 或者 每个机器申请一个对应IP的证书

---

生成 etcd 证书和私钥：

cfssl gencert -ca=ca.pem \-ca-key=ca-key.pem \-config=ca-config.json \-profile=kubernetes etcd-csr.json | cfssljson -bare etcdls etcd*mkdir -p /etc/etcd/ssl
cp etcd.pem etcd-key.pem  ca.pem /etc/etcd/ssl/
##其他node
rm -rf /etc/etcd/ssl/*
scp -r /etc/etcd/ssl root@10.129.6.211:/etc/etcd/scp -r root@10.129.6.211:/root/k8s/etcd/etcd-v3.3.0-rc.1-linux-amd64.tar.gz /root

将生成好的etcd.pem和etcd-key.pem以及ca.pem三个文件拷贝到目标主机的/etc/etcd/ssl目录下。

下载二进制安装文件

到 https://github.com/coreos/etcd/releases 页面下载最新版本的二进制文件：

wget http://github.com/coreos/etcd/releases/download/v3.1.10/etcd-v3.1.10-linux-amd64.tar.gztar -xvf etcd-v3.1.10-linux-amd64.tar.gzmv etcd-v3.1.10-linux-amd64/etcd* /usr/local/bin

创建 etcd 的 systemd unit 文件

mkdir -p /var/lib/etcd  # 必须先创建工作目录
cat > etcd.service <<EOF
[Unit]
Description=Etcd Server
After=network.target
After=network-online.target
Wants=network-online.target
Documentation=https://github.com/coreos[Service]
Type=notify
WorkingDirectory=/var/lib/etcd/
ExecStart=/usr/local/bin/etcd \\--name=${NODE_NAME} \\--cert-file=/etc/etcd/ssl/etcd.pem \\--key-file=/etc/etcd/ssl/etcd-key.pem \\--peer-cert-file=/etc/etcd/ssl/etcd.pem \\--peer-key-file=/etc/etcd/ssl/etcd-key.pem \\--trusted-ca-file=/etc/etcd/ssl/ca.pem \\--peer-trusted-ca-file=/etc/etcd/ssl/ca.pem \\--initial-advertise-peer-urls=https://${NODE_IP}:2380 \\--listen-peer-urls=https://${NODE_IP}:2380 \\--listen-client-urls=https://${NODE_IP}:2379,http://127.0.0.1:2379 \\--advertise-client-urls=https://${NODE_IP}:2379 \\--initial-cluster-token=etcd-cluster-0 \\--initial-cluster=${ETCD_NODES} \\--initial-cluster-state=new \\--data-dir=/var/lib/etcd
Restart=on-failure
RestartSec=5
LimitNOFILE=65536[Install]
WantedBy=multi-user.target
EOF

• 指定 etcd 的工作目录和数据目录为 /var/lib/etcd，需在启动服务前创建这个目录；
• 为了保证通信安全，需要指定 etcd 的公私钥(cert-file和key-file)、Peers 通信的公私钥和 CA 证书(peer-cert-file、peer-key-file、peer-trusted-ca-file)、客户端的CA证书（trusted-ca-file）；
• --initial-cluster-state 值为 new 时，--name 的参数值必须位于 --initial-cluster 列表中；

启动 etcd 服务

mv etcd.service /etc/systemd/system/
systemctl daemon-reload
systemctl enable etcd
systemctl start etcd
systemctl status etcd

验证服务

 etcdctl \--endpoints=https://${NODE_IP}:2379  \--ca-file=/etc/etcd/ssl/ca.pem \--cert-file=/etc/etcd/ssl/etcd.pem \--key-file=/etc/etcd/ssl/etcd-key.pem \cluster-health

预期结果：

[root@node02 ~]# etcdctl   --endpoints=https://${NODE_IP}:2379    --ca-file=/etc/etcd/ssl/ca.pem   --cert-file=/etc/etcd/ssl/etcd.pem   --key-file=/etc/etcd/ssl/etcd-key.pem   cluster-health
member 18699a64c36a7e7b is healthy: got healthy result from https://10.129.6.213:2379
member 5dbd6a0b2678c36d is healthy: got healthy result from https://10.129.6.211:2379
member 6b1bf02f85a9e68f is healthy: got healthy result from https://10.129.6.212:2379
cluster is healthy

若有失败 或 重新配置

systemctl stop etcdrm -Rf /var/lib/etcdrm -Rf /var/lib/etcd-clustermkdir -p /var/lib/etcdsystemctl start etcd

k8s 安装

提取k8s rpm 包

• 默认由于某某出海问题
• 我们离线导入下rpm 仓库
• 安装官方YUM 仓库

cat <<EOF > /etc/yum.repos.d/kubernetes.repo
[kubernetes]
name=Kubernetes
baseurl=http://yum.kubernetes.io/repos/kubernetes-el7-x86_64
enabled=1
gpgcheck=1
repo_gpgcheck=0
gpgkey=https://packages.cloud.google.com/yum/doc/yum-key.gpghttps://packages.cloud.google.com/yum/doc/rpm-package-key.gpg
EOF

安装kubeadm kubectl cni

• 下载镜像(自行搬×××先获取rpm)

mkdir -p  /root/k8s/rpm
cd /root/k8s/rpm#安装同步工具
yum install -y yum-utils#同步本地镜像
yumdownloader kubelet kubeadm kubectl kubernetes-cni docker
scp root@10.129.6.224:/root/k8s/rpm/*  /root/k8s/rpm

• 离线安装

mkdir -p  /root/k8s/rpm
scp root@10.129.6.211:/root/k8s/rpm/*  /root/k8s/rpm
yum install  /root/k8s/rpm/*.rpm -y

• 启动k8s

#restart
systemctl enable docker && systemctl restart docker
systemctl enable kubelet && systemctl restart kubelet

镜像获取方法

• 加速器获取 gcr.io k8s镜像 ，导出，导入镜像 或 上传本地仓库

#国内可以使用daocloud加速器下载相关镜像，然后通过docker save、docker load把本地下载的镜像放到kubernetes集群的所在机器上，daocloud加速器链接如下：
https://www.daocloud.io/mirror#accelerator-doc#pull 获取
docker pull gcr.io/google_containers/kube-proxy-amd64:v1.9.0#导出
mkdir -p docker-images
docker save -o docker-images/kube-proxy-amd64  gcr.io/google_containers/kube-proxy-amd64:v1.9.0#导入
docker load -i /root/kubeadm-ha/docker-images/kube-proxy-amd64

• 代理或***获取 gcr.io k8s镜 ,导出，导入镜像 或 上传本地仓库

自谋生路，天机屋漏

kubelet 指定本地镜像

kubelet 修改 配置以使用本地自定义pause镜像
devhub.beisencorp.com/google_containers/pause-amd64:3.0 替换你的环境镜像

cat > /etc/systemd/system/kubelet.service.d/20-pod-infra-image.conf <<EOF
[Service]
Environment="KUBELET_EXTRA_ARGS=--pod-infra-container-image=devhub.beisencorp.com/google_containers/pause-amd64:3.0"
EOF
systemctl daemon-reload
systemctl restart kubelet

Kubeadm Init 初始化

• 我们使用config 模板方式来初始化集群，便于我们指定etcd 集群
• devhub.beisencorp.com 使我们的 测试镜像仓库 可以改成自己或者手动导入每个机器镜像

cat <<EOF > config.yaml
apiVersion: kubeadm.k8s.io/v1alpha1
kind: MasterConfiguration
etcd:endpoints:- https://10.129.6.211:2379- https://10.129.6.212:2379- https://10.129.6.213:2379caFile: /etc/etcd/ssl/ca.pem certFile: /etc/etcd/ssl/etcd.pem keyFile: /etc/etcd/ssl/etcd-key.pemdataDir: /var/lib/etcd
networking:podSubnet: 10.244.0.0/16
kubernetesVersion: 1.9.0
api:advertiseAddress: "10.129.6.220"
token: "b99a00.a144ef80536d4344"
tokenTTL: "0s"
apiServerCertSANs:
- etcd-host1
- etcd-host2
- etcd-host3
- 10.129.6.211
- 10.129.6.212
- 10.129.6.213
- 10.129.6.220
featureGates:CoreDNS: true
imageRepository: "devhub.beisencorp.com/google_containers"
EOF

• 初始化集群

kubeadm init --config config.yaml 

• 结果

To start using your cluster, you need to run the following as a regular user:mkdir -p $HOME/.kubesudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/configsudo chown $(id -u):$(id -g) $HOME/.kube/config
as root:kubeadm join --token b99a00.a144ef80536d4344 10.129.6.220:6443 --discovery-token-ca-cert-hash sha256:ebc2f64e9bcb14639f26db90288b988c90efc43828829c557b6b66bbe6d68dfa

• 查看node

[root@etcd-host1 k8s]# kubectl get node
NAME         STATUS    ROLES     AGE       VERSION
etcd-host1   noReady     master    5h        v1.9.0
[root@etcd-host1 k8s]# kubectl get cs
NAME                 STATUS    MESSAGE              ERROR
scheduler            Healthy   ok
controller-manager   Healthy   ok
etcd-1               Healthy   {"health": "true"}
etcd-2               Healthy   {"health": "true"}
etcd-0               Healthy   {"health": "true"}  

• 问题记录

如果使用kubeadm初始化集群，启动过程可能会卡在以下位置，那么可能是因为cgroup-driver参数与docker的不一致引起
[apiclient] Created API client, waiting for the control plane to become ready
journalctl -t kubelet -S '2017-06-08'查看日志，发现如下错误
error: failed to run Kubelet: failed to create kubelet: misconfiguration: kubelet cgroup driver: "systemd"
需要修改KUBELET_CGROUP_ARGS=--cgroup-driver=systemd为KUBELET_CGROUP_ARGS=--cgroup-driver=cgroupfsvi /etc/systemd/system/kubelet.service.d/10-kubeadm.conf
#Environment="KUBELET_CGROUP_ARGS=--cgroup-driver=systemd"
Environment="KUBELET_CGROUP_ARGS=--cgroup-driver=cgroupfs"systemctl daemon-reload && systemctl restart kubelet

安装网络组件 podnetwork

• 我们选用kube-router

wget https://github.com/cloudnativelabs/kube-router/blob/master/daemonset/kubeadm-kuberouter.yamlkubectl apply -f kubeadm-kuberouter.yaml

• 结果

[root@etcd-host1 k8s]# kubectl get po --all-namespaces
NAMESPACE     NAME                                 READY     STATUS    RESTARTS   AGE
kube-system   coredns-546545bc84-zc5dx             1/1       Running   0          6h
kube-system   kube-apiserver-etcd-host1            1/1       Running   0          6h
kube-system   kube-controller-manager-etcd-host1   1/1       Running   0          6h
kube-system   kube-proxy-pfj7x                     1/1       Running   0          6h
kube-system   kube-router-858b7                    1/1       Running   0          37m
kube-system   kube-scheduler-etcd-host1            1/1       Running   0          6h
[root@etcd-host1 k8s]# 

部署其他Master 节点

• 拷贝master01 配置 master02 master03

#拷贝pki 证书
mkdir -p /etc/kubernetes/pki
scp -r root@10.129.6.211:/etc/kubernetes/pki /etc/kubernetes#拷贝初始化配置
scp -r root@10.129.6.211://root/k8s/config.yaml /etc/kubernetes/config.yaml

• 初始化 master02 master03

#初始化
kubeadm init --config /etc/kubernetes/config.yaml 

部署成功 验证结果

为了测试我们把master 设置为 可部署role

默认情况下，为了保证master的安全，master是不会被调度到app的。你可以取消这个限制通过输入：

kubectl taint nodes --all node-role.kubernetes.io/master-

---

录制终端验证 结果

-验证
[zeming@etcd-host1 k8s]$ kubectl get node
NAME         STATUS    ROLES     AGE       VERSION
etcd-host1   Ready     master    6h        v1.9.0
etcd-host2   Ready     master    5m        v1.9.0
etcd-host3   Ready     master    49s       v1.9.0
[zeming@etcd-host1 k8s]$ kubectl get po --all-namespaces
NAMESPACE     NAME                                 READY     STATUS    RESTARTS   AGE
default       nginx01-d87b4fd74-2445l              1/1       Running   0          1h
default       nginx01-d87b4fd74-7966r              1/1       Running   0          1h
default       nginx01-d87b4fd74-rcbhw              1/1       Running   0          1h
kube-system   coredns-546545bc84-zc5dx             1/1       Running   0          3d
kube-system   kube-apiserver-etcd-host1            1/1       Running   0          3d
kube-system   kube-apiserver-etcd-host2            1/1       Running   0          3d
kube-system   kube-apiserver-etcd-host3            1/1       Running   0          3d
kube-system   kube-controller-manager-etcd-host1   1/1       Running   0          3d
kube-system   kube-controller-manager-etcd-host2   1/1       Running   0          3d
kube-system   kube-controller-manager-etcd-host3   1/1       Running   0          3d
kube-system   kube-proxy-gk95d                     1/1       Running   0          3d
kube-system   kube-proxy-mrzbq                     1/1       Running   0          3d
kube-system   kube-proxy-pfj7x                     1/1       Running   0          3d
kube-system   kube-router-bbgpq                    1/1       Running   0          3h
kube-system   kube-router-v2jbh                    1/1       Running   0          3h
kube-system   kube-router-w4cbb                    1/1       Running   0          3h
kube-system   kube-scheduler-etcd-host1            1/1       Running   0          3d
kube-system   kube-scheduler-etcd-host2            1/1       Running   0          3d
kube-system   kube-scheduler-etcd-host3            1/1       Running   0          3d
[zeming@etcd-host1 k8s]$

主备测试

• 关闭 主节点 master01 观察切换到 master02 机器
• master03 一直不管获取node信息 测试高可用

while true; do  sleep 1; kubectl get node;date; done

观察主备VIP切换过程

#观察当Master01主节点关闭后，被节点VIP状态  BACKUP  切换到 MASTER
[root@etcd-host2 net.d]# systemctl status keepalived
● keepalived.service - LVS and VRRP High Availability MonitorLoaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)Active: active (running) since Tue 2018-01-22 13:54:17 CST; 21s agoJan 22 13:54:17 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) Transition to MASTER STATE
Jan 22 13:54:17 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) Received advert with higher priority 120, ours 110
Jan 22 13:54:17 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) Entering BACKUP STATE#切换到 MASTER
[root@etcd-host2 net.d]# systemctl status keepalived
● keepalived.service - LVS and VRRP High Availability MonitorLoaded: loaded (/usr/lib/systemd/system/keepalived.service; enabled; vendor preset: disabled)Active: active (running) since Tue 2018-01-22 13:54:17 CST; 4min 6s agoJan 22 14:03:02 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) Transition to MASTER STATE
Jan 22 14:03:03 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) Entering MASTER STATE
Jan 22 14:03:03 etcd-host2 Keepalived_vrrp[15908]: VRRP_Instance(VI_1) setting protocol VIPs.
Jan 22 14:03:03 etcd-host2 Keepalived_vrrp[15908]: Sending gratuitous ARP on ens32 for 10.129.6.220

验证集群高可用

#观察 master01 关机后状态变成NotReady
[root@etcd-host3 ~]# while true; do  sleep 1; kubectl get node;date; done
Tue Jan 22 14:03:16 CST 2018
NAME         STATUS    ROLES     AGE       VERSION
etcd-host1   Ready     master    19m       v1.9.0
etcd-host2   Ready     master    3d        v1.9.0
etcd-host3   Ready     master    3d        v1.9.0
Tue Jan 22 14:03:17 CST 2018
NAME         STATUS     ROLES     AGE       VERSION
etcd-host1   NotReady   master    19m       v1.9.0
etcd-host2   Ready      master    3d        v1.9.0
etcd-host3   Ready      master    3d        v1.9.0#恢复Master主节点后，出现VIP偏移过来，api恢复
The connection to the server 10.129.6.220:6443 was refused - did you specify the right host or port?
Tue Jan 23 14:14:05 CST 2018
The connection to the server 10.129.6.220:6443 was refused - did you specify the right host or port?
Tue Jan 23 14:14:07 CST 2018
Tue Jan 23 14:14:18 CST 2018
NAME         STATUS     ROLES     AGE       VERSION
etcd-host1   NotReady   master    29m       v1.9.0
etcd-host2   Ready      master    3d        v1.9.0
etcd-host3   Ready      master    3d        v1.9.0
Tue Jan 23 14:14:20 CST 2018
NAME         STATUS    ROLES     AGE       VERSION
etcd-host1   Ready     master    29m       v1.9.0
etcd-host2   Ready     master    3d        v1.9.0
etcd-host3   Ready     master    3d        v1.9.0

---

参观文档

#k8s 官方文档
https://kubernetes.io/docs/reference/setup-tools/kubeadm/kubeadm-init/
#kubeadm ha 项目文档
https://github.com/indiketa/kubeadm-ha
https://github.com/cookeem/kubeadm-ha/blob/master/README_CN.md
https://medium.com/@bambash/ha-kubernetes-cluster-via-kubeadm-b2133360b198
#kubespray 之前的kargo ansible项目
https://github.com/kubernetes-incubator/kubespray/blob/master/docs/ha-mode.md
#若有问题或转载请注明出处 By Zeming

http://xuzeming.top/2018/01/19/K8s-1-9%E5%AE%9E%E8%B7%B5-Kubeadm-HA-1-9-%E9%AB%98%E5%8F%AF%E7%94%A8-%E9%9B%86%E7%BE%A4-%E6%9C%AC%E5%9C%B0%E7%A6%BB%E7%BA%BF%E9%83%A8%E7%BD%B2/