Kubernetes/部署文档/Rancher/Rke1部署Kubernetes集群.md
offends 7a2f41e7d6
All checks were successful
continuous-integration/drone Build is passing
synchronization
2024-08-07 18:54:39 +08:00

9.2 KiB
Raw Blame History

本文作者:丁辉

Rke1部署Kubernetes集群

RKE1文档

Rancher中文文档

节点名称 IP Kubernetes角色
k8s-master-1,Rke管理 192.168.1.10 controlplane,etcd,worker
k8s-master-2 192.168.1.20 controlplane,etcd,worker
k8s-master-3 192.168.1.30 controlplane,etcd,worker

环境准备

!!!每次部署都写挺麻烦的索性都放在一个文件内了请查看 Kubernetes基础环境准备 ,请按照此文档初始化环境

所有节点执行

  1. 配置 SSH

    sed -i 's/#AllowTcpForwarding yes/AllowTcpForwarding yes/g' /etc/ssh/sshd_config
    

    重启 SSH

    systemctl restart sshd
    
  2. 将用户添加到 docker 组

    groupadd docker
    useradd -m docker -g docker
    

    使用其他用户

    useradd rke # 创建用户
    usermod -aG docker rke  将rke用户加入docker组
    
  3. 配置 docker 用户免密登录

    mkdir -p /home/docker/.ssh/
    touch /home/docker/.ssh/authorized_keys
    chmod 700 /home/docker/.ssh/
    chown -R docker.docker /home/docker/.ssh/
    chmod 600 /home/docker/.ssh/authorized_keys
    

Rke管理节点执行

  1. 生成密钥

    ssh-keygen -t rsa -N "" -f ~/.ssh/id_rsa -q
    
  2. 查看主节点密钥

    密钥需要到 RKE 初始化节点上获取,所有节点都是用此密钥

    cat ~/.ssh/id_rsa.pub
    

所有节点执行

  1. 粘贴密钥内容到此文件内(提示所有节点粘贴Rke管理节密钥)

    vi /home/docker/.ssh/authorized_keys
    
  2. 验证是否可以免密登录

    ssh docker@192.168.1.10
    

Docker安装

  1. Docker安装

    curl https://releases.rancher.com/install-docker/20.10.sh | sh
    

    传递参数使用国内源

    curl -fsSL https://releases.rancher.com/install-docker/20.10.sh | sh -s -- --mirror Aliyun
    
  2. 启动 Docker

    systemctl enable docker
    systemctl start docker
    

安装并初始化Rke

RKE二进制文件

  1. 下载 RKE 二进制文件,并添加到可执行路径下

    wget https://github.com/rancher/rke/releases/download/v1.4.10/rke_linux-amd64
    
  2. 授权

    chmod 777 rke_linux-amd64 && mv rke_linux-amd64  /usr/local/bin/rke
    

方法一 (不推荐怪麻烦的请看"方法二")

如果 Number of Hosts 填的是多节点则会提示输入多次节点信息

rke config --name cluster.yml
[+] Cluster Level SSH Private Key Path [~/.ssh/id_rsa]: #默认回车
[+] Number of Hosts [1]: #节点数量
[+] SSH Address of host (1) [none]: 192.168.1.10 #节点IP地址
[+] SSH Port of host (1) [22]: #默认回车
[+] SSH Private Key Path of host (192.168.1.10) [none]: #默认回车
[-] You have entered empty SSH key path, trying fetch from SSH key parameter
[+] SSH Private Key of host (192.168.1.10) [none]: #默认回车
[-] You have entered empty SSH key, defaulting to cluster level SSH key: ~/.ssh/id_rsa
[+] SSH User of host (192.168.1.10) [ubuntu]: docker #SSH用户
[+] Is host (192.168.1.10) a Control Plane host (y/n)? [y]: y #是否为控制节点
[+] Is host (192.168.1.10) a Worker host (y/n)? [n]: y #是否为计算节点
[+] Is host (192.168.1.10) an etcd host (y/n)? [n]: y #是否为etcd节点
[+] Override Hostname of host (192.168.1.10) [none]: #默认回车
[+] Internal IP of host (192.168.1.10) [none]: 192.168.1.10 #主机内部IP
[+] Docker socket path on host (192.168.1.10) [/var/run/docker.sock]: #默认回车
[+] Network Plugin Type (flannel, calico, weave, canal, aci) [canal]: flannel #选择网络插件类型
[+] Authentication Strategy [x509]: #默认回车
[+] Authorization Mode (rbac, none) [rbac]: #默认回车
[+] Kubernetes Docker image [rancher/hyperkube:v1.26.8-rancher1]: #选择 k8s 版本
[+] Cluster domain [cluster.local]: #集群域
[+] Service Cluster IP Range [10.43.0.0/16]: #服务集群IP范围
[+] Enable PodSecurityPolicy [n]: #默认回车
[+] Cluster Network CIDR [10.42.0.0/16]: #集群网络CIDR
[+] Cluster DNS Service IP [10.43.0.10]: #集群DNS服务IP
[+] Add addon manifest URLs or YAML files [no]: #默认回车

基础参数修改

sed -i '/^ingress:$/,/^  provider:/ s/provider: ""/provider: "none"/' cluster.yml

方法二

  1. 生成初始文件

    rke config --empty --name cluster.yml
    
  2. 按需要修改 address 参数

    多节点则写多个 address 并通过调整 role 指定节点属性

    nodes:
    - address: 192.168.1.10
      port: "22"
      internal_address: 192.168.1.10
      role:
      - controlplane #管理
      - worker #计算
      - etcd #etcd节点
      hostname_override: ""
      user: docker
      docker_socket: /var/run/docker.sock
      ssh_key: ""
      ssh_key_path: ~/.ssh/id_rsa
      ssh_cert: ""
      ssh_cert_path: ""
      labels: {}
      taints: []
    # 格式一样此处省略 20,30 节点配置 ...
    services:
    ...
    
  3. 基础参数修改

    sed -i 's/service_cluster_ip_range: ""/service_cluster_ip_range: 10.43.0.0\/16/' cluster.yml
    sed -i 's/cluster_cidr: ""/cluster_cidr: 10.42.0.0\/16/' cluster.yml
    sed -i 's/cluster_domain: ""/cluster_domain: cluster.local/' cluster.yml
    sed -i 's/cluster_dns_server: ""/cluster_dns_server: 10.43.0.10/' cluster.yml
    sed -i 's/plugin: ""/plugin: flannel/' cluster.yml
    sed -i 's/strategy: ""/strategy: x509/' cluster.yml
    sed -i 's/^\s*mode: ""$/  mode: rbac/' cluster.yml
    sed -i '/^ingress:$/,/^  provider:/ s/provider: ""/provider: "none"/' cluster.yml
    sed -i '/^[^ ]/ s/ssh_key_path: ""/ssh_key_path: ~\/.ssh\/id_rsa/g' cluster.yml
    sed -i '0,/^\s*ssh_key_path: ""$/{s,^\s*ssh_key_path: ""$,  ssh_key_path: ~/.ssh/id_rsa,}' cluster.yml
    

初始化 Kubernetes 集群

rke up
  • 禁用 metrics-server 组件

    sed -i '/^monitoring:$/,/^  provider:/ s/provider: ""/provider: "none"/' cluster.yml
    
  • 调整节点端口范围

    默认端口范围30000-32767

    sed -i 's/service_node_port_range: ""/service_node_port_range: "10000-30000"/' cluster.yml
    
  • 关闭 Docker 版本检测

    sed -i 's/ignore_docker_version: null/ignore_docker_version: true/' cluster.yml
    
  • 调整部署版本

    • 查看当前 RKE 支持的Kubernetes版本
    rke config --list-version --all
    
    • 替换版本
    sed -i 's/kubernetes_version: ""/kubernetes_version: "v1.24.17-rancher1-1"/' cluster.yml
    
  • 更新集群

    rke up --update-only
    

安装 kubectl

Kubectl二进制文件

  1. 下载 kubectl

    curl -LO https://dl.k8s.io/release/v1.26.8/bin/linux/amd64/kubectl
    
  2. 授权

    chmod 777 kubectl && mv kubectl /usr/local/bin/
    
  3. 添加 kubctl 文件

    mkdir ~/.kube && cp kube_config_cluster.yml ~/.kube/config && chmod 600 ~/.kube/config
    
  4. 验证

    kubectl get node
    

    本文中没有禁用 monitoring 所以也可以使用 kubectl top node 测试

卸载

  1. 卸载 RKE 集群

    rke remove
    
  2. 清理残余容器

    for i in $(docker ps -a | grep rancher | awk '{print $1}');do docker rm -f $i;done
    for i in $(docker ps -a | grep rke | awk '{print $1}');do docker rm -f $i;done
    
  3. 清除 Docker 引擎的废弃资源和缓存

    docker system prune --all
    
  4. 卸载挂载

    mount | grep /var/lib/kubelet/pods/ | awk '{print $1}' | xargs umount -l
    
  5. 删除持久化目录

    rm -rf /var/lib/kubelet/
    rm -rf /run/flannel/
    

备份和恢复

非常重要,他奶奶的吃大亏了

创建一次性快照

RKE 会将节点快照保存在 /opt/rke/etcd-snapshots 路径下

rke etcd snapshot-save --config cluster.yml --name <快照名称>

恢复集群

rke etcd snapshot-restore --config cluster.yml --name <快照名称>

恢复 Rke配置文件

准备依赖 jq

  • Centos

    yum install jq -y
    
  • Ubuntu

    apt install jq -y
    
  • 恢复 Kubectl 配置文件

    修改 --master-ip= 为任意 K8S Master节点IP

    curl -sfL https://gitee.com/offends/Kubernetes/raw/main/File/Shell/restore-rkestate-kubeconfig.sh | bash -s -- --master-ip=<K8S Master节点IP>
    
  • 恢复 rkestate 状态文件

    • 通过本地 kubectl 找回

      kubectl get configmap -n kube-system full-cluster-state -o json | jq -r .data.\"full-cluster-state\" | jq -r . > cluster.rkestate
      
    • 通过 master 节点找回

      curl -sfL https://gitee.com/offends/Kubernetes/raw/main/File/Shell/restore-rkestate-config.sh | bash -s