前提：
环境centeos 7
可以上外网，因为所有的工具都是通过外网获取的
在master和node上执行
解决镜像慢的问题
*根据具体情况进行修改
sudo mkdir -p /etc/docker
sudo tee /etc/docker/daemon.json <<-'EOF'
{
"registry-mirrors": ["https://*.mirror.aliyuncs.com"]
}
EOF
sudo systemctl daemon-reload
sudo systemctl restart docker
master节点执行
配置docker yum源（自行配置）
这个repo主要用来装docker
yum install -y docker
systemctl start docker
配置kubernetes的yum源
ali_kube.repo
[kubernetes_ali]
name=Kubernetes
baseurl=https://mirrors.aliyun.com/kubernetes/yum/repos/kubernetes-el7-x86_64
enabled=1
gpgcheck=0
这个repo用来安装kubeadm和kubelet
目前是v1.9.6的kubeadm
可以使用 kubeadm version 查看版本
yum install -y kubeadm kubelet
（根据版本自己拼装yum install -y kubeadm-1.9.6-0.x86_64 kubelet-1.9.6-0.x86_64）
预先下载相关的镜像
如果不事先下好，会默认到国外的网站去pulldocker镜像，这样就会失败的
下载好之后还要把docker镜像名打个tag，不然kubeadm是不识别的
docker_pull 文件
#
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/pause-amd64:3.0
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/kube-apiserver-amd64:v1.9.6
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/kube-controller-manager-amd64:v1.9.6
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/kube-scheduler-amd64:v1.9.6
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/kube-proxy-amd64:v1.9.6
#docker pull registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.10
docker pull registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.11
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-sidecar-amd64:1.14.7
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-kube-dns-amd64:1.14.7
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-dnsmasq-nanny-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/pause-amd64:3.0 k8s.gcr.io/pause-amd64:3.0
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-apiserver-amd64:v1.9.6 k8s.gcr.io/kube-apiserver-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-controller-manager-amd64:v1.9.6 k8s.gcr.io/kube-controller-manager-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-scheduler-amd64:v1.9.6 k8s.gcr.io/kube-scheduler-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-proxy-amd64:v1.9.6 k8s.gcr.io/kube-proxy-amd64:v1.9.6
#docker tag registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.10 k8s.gcr.io/etcd-amd64:3.1.10
docker tag registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.11 k8s.gcr.io/etcd-amd64:3.1.11
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-sidecar-amd64:1.14.7 k8s.gcr.io/k8s-dns-sidecar-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-kube-dns-amd64:1.14.7 k8s.gcr.io/k8s-dns-kube-dns-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-dnsmasq-nanny-amd64:1.14.7 k8s.gcr.io/k8s-dns-dnsmasq-nanny-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/pause-amd64:3.0 gcr.io/google_containers/pause-amd64:3.0
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-apiserver-amd64:v1.9.6 gcr.io/google_containers/kube-apiserver-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-controller-manager-amd64:v1.9.6 gcr.io/google_containers/kube-controller-manager-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-scheduler-amd64:v1.9.6 gcr.io/google_containers/kube-scheduler-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-proxy-amd64:v1.9.6 gcr.io/google_containers/kube-proxy-amd64:v1.9.6
#docker tag registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.10 gcr.io/google_containers/etcd-amd64:3.1.10
docker tag registry.cn-hangzhou.aliyuncs.com/capfgd/etcd-amd64:3.1.11 gcr.io/google_containers/etcd-amd64:3.1.11
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-sidecar-amd64:1.14.7 gcr.io/google_containers/k8s-dns-sidecar-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-kube-dns-amd64:1.14.7 gcr.io/google_containers/k8s-dns-kube-dns-amd64:1.14.7
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/k8s-dns-dnsmasq-nanny-amd64:1.14.7 gcr.io/google_containers/k8s-dns-dnsmasq-nanny-amd64:1.14.7
kubeadm init --kubernetes-version=1.9.6
使用这个命令初始化kubernetes集群的master节点
安装成功之后执行
useradd k8s
mkdir -p /home/k8s/.kube
cp -i /etc/kubernetes/admin.conf /home/k8s/.kube/config
vim /home/k8s/.kube/config
ll /home/k8s/.kube/config
chown k8s:k8s /home/k8s/.kube/config
ll /home/k8s/.kube/config
遇到的问题
1.如果docker镜像手动拉取太慢，可以使用阿里云的镜像加速，网站https://cr.console.aliyun.com
并且可以在这里搜索可用的镜像以及对应的版本
将自己私有的加速网址加到docker的启动参数中，OPTIONS='--registry-mirror=http://*****'
2.使用kubeadm 初始化集群的时候会出现很多的错误，这个时候需要观察日志来排查到底是哪里出了问题。需要观察的日志是kubelet和docker的，默认是在/var/log/message文件中
tail -f /var/log/messages | grep kubelet
tail -f /var/log/messages | grep docker
如果kubelet日志碰到 network plugin is not ready: cni config uninitialized
将/etc/systemd/system/kubelet.service.d/10-kubeadm.conf中$KUBELET_NETWORK_ARGS注释
刚开始kubelet会出现类似
Mar 28 15:42:02 cloudboot kubelet: E0328 15:42:02.796341 15751 reflector.go:205] k8s.io/kubernetes/pkg/kubelet/kubelet.go:480: Failed to list *v1.Node: Get https://172.16.3.144:6443/api/v1/nodes?fieldSelector=metadata.name%3Dcloudboot&limit=500&resourceVersion=0: dial tcp 172.16.3.144:6443: getsockopt: connection refused
这个属于正常的，之后会消失
3.不同kubeadm找的docker镜像名和版本都是不一样的，就连官网说的那个版本https://kubernetes.io/docs/reference/setup-tools/kubeadm/kubeadm-init/#running-kubeadm-without-an-internet-connection 里面的名称和版本都有错的，以实际日志中给出的信息为准
工作节点
Ubuntu
apt install curl
apt-get install -y apt-transport-https
apt install docker.io
curl -s[https://mirrors.aliyun.com/kubernetes/apt/doc/](https://mirrors.aliyun.com/kubernetes/apt/doc/)apt-key.gpg | apt-key add -
cat < /etc/apt/sources.list.d/kubernetes.list
debhttps://mirrors.aliyun.com/kubernetes/apt/kubernetes-xenial main
EOF
apt-get update
apt-get install -y kubelet kubeadm kubectl kubernetes-cni
(指定版本 apt-get install -y kubelet=1.9.6-00 kubeadm=1.9.6-00 kubectl=1.9.6-00 kubernetes-cni=0.6.0-00)
systemctl start docker && systemctl enable docker
redhat
见master节点yum仓库的配置
如果为了上网而执行了http_proxy或https_proxy 加入节点前一定要unset这些代理不然会出现连接forbidden的问题
为docker设置代理，不用在系统变量里设置
mkdir /etc/systemd/system/docker.service.d
cat /etc/systemd/system/docker.service.d/http-proxy.conf
[Service]
Environment="HTTP_PROXY=http://172.16.3.11:3128"
Environment="HTTPS_PROXY=http://172.16.3.11:3128"
systemctl daemon-reload
systemctl restart docker
加入节点
kubeadm join --token 4b1a19.d0e8c2b47a36e6d7 172.16.38.210:6443 --discovery-token-ca-cert-hash sha256:4beb4dc70f098f69a0fd4d85e61c9dc852ff1bd8e855a999d00ae7f87e2e54c8
上面的命令是master节点安装成功后给出的，如果是第二次加入token需要重新生成(在master节点上执行 kubeadm token create)
替换上面的token
su - k8s
kubectl get nodes
NAME STATUS ROLES AGE VERSION
cloudboot Ready master 2h v1.9.6
ubuntu-kvm NotReady 5m v1.9.6
Ubuntu系统的日志在
tailf /var/log/syslog | grep kubelet
通过日志可以看到还是需要装一些东西的（可能会需要disable network args）
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/pause-amd64:3.0
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/pause-amd64:3.0 gcr.io/google_containers/pause-amd64:3.0
docker pull registry.cn-shenzhen.aliyuncs.com/eaglering/kube-proxy-amd64:v1.9.6
docker tag registry.cn-shenzhen.aliyuncs.com/eaglering/kube-proxy-amd64:v1.9.6 gcr.io/google_containers/kube-proxy-amd64:v1.9.6
在master节点上执行 (只执行一次)
kubectl apply -f "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')"
当前pods的运行状态
kubectl get pods --all-namespaces
需要都运行正常（running）
kubectl --namespace=kube-system describe pod weave-net-x7k59
查看具体哪个pods的出错信息
因为weave也需要pull镜像，并且在master和node上都装不上，手动拉取吧
weaveworks/weave-kube:2.2.1
weaveworks/weave-npc:2.2.1
docker pull weaveworks/weave-kube:2.2.1
docker pull weaveworks/weave-npc:2.2.1
如果需要清空kubernetes安装环境
在每个节点（master和node）上kubeadm reset重新安装，master节点kubeadm init
