K8s(v1.25.1) 高可用集群(3 Master + 5 Node) Ansible 剧本部署(CRI使用docker,cri-docker)
写在前面
- 分享一个 k8s 高可用集群部署的 Ansible 剧本
- 以及涉及到的一些工具的安装
- 博文内容涉及：
- 从零开始 一个 k8s 高可用 集群部署 Ansible剧本编写，
- 编写后搭建 k8s 高可用 集群
- 一些集群常用的 监控，备份工具安装，包括：
cadvisor监控工具部署metrics-server监控工具部署Ingress—nginxIngress 控制器部署Metallb软 LoadBalancer 部署local-path-storage基于本地存储的SC 分配器部署prometheus监控工具部署ETCD快照备份定时任务编写运行Velero集群容灾备份工具部署
- 理解不足小伙伴帮忙指正
我所渴求的，無非是將心中脫穎語出的本性付諸生活，為何竟如此艱難呢 ------赫尔曼·黑塞《德米安》
部署完的 集群 node 信息查看,感觉版本有点高，生产不太建议安装这么高的版本，有些开源工具都安装不了，各种问问题…
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$kubectl get nodesNAME STATUS ROLES AGE VERSIONvms100.liruilongs.github.io Ready control-plane 10d v1.25.1vms101.liruilongs.github.io Ready control-plane 10d v1.25.1vms102.liruilongs.github.io Ready control-plane 10d v1.25.1vms103.liruilongs.github.io Ready <none> 10d v1.25.1vms105.liruilongs.github.io Ready <none> 10d v1.25.1vms106.liruilongs.github.io Ready <none> 10d v1.25.1vms107.liruilongs.github.io Ready <none> 10d v1.25.1vms108.liruilongs.github.io Ready <none> 10d v1.25.1部署涉及机器
- master1: 192.168.26.100
- master2: 192.168.26.101
- master3: 192.168.26.102
- work Node1: 192.168.26.103
- work Node2: 192.168.26.105
- work Node3: 192.168.26.106
- work Node4: 192.168.26.107
- work Node5: 192.168.26.108
哈 104 不吉利，跳过去了…
集群部署
部署拓扑
下图为 k8s 官网文档中 HA 拓扑图，这里使用下面的拓扑方式部署
堆叠(Stacked)HA 集群拓扑， 其中 etcd 分布式数据存储集群堆叠在 kubeadm 管理的控制平面节点上，作为控制平面的一个组件运行。
部署后的 etcd 分布
┌──[root@vms100.liruilongs.github.io]-[~/ansible/kubescape]└─$ETCDCTL_API=3 etcdctl --endpoints https://127.0.0.1:2379 --cert="/etc/kubernetes/pki/etcd/server.crt" --key="/etc/kubernetes/pki/etcd/server.key" --cacert="/etc/kubernetes/pki/etcd/ca.crt" member list -w table+------------------+---------+-----------------------------+-----------------------------+-----------------------------+| ID | STATUS | NAME | PEER ADDRS | CLIENT ADDRS |+------------------+---------+-----------------------------+-----------------------------+-----------------------------+| ee392e5273e89e2 | started | vms100.liruilongs.github.io | https://192.168.26.100:2380 | https://192.168.26.100:2379 || 11486647d7f3a17b | started | vms102.liruilongs.github.io | https://192.168.26.102:2380 | https://192.168.26.102:2379 || e00e3877df8f76f4 | started | vms101.liruilongs.github.io | https://192.168.26.101:2380 | https://192.168.26.101:2379 |+------------------+---------+-----------------------------+-----------------------------+-----------------------------+每个控制平面节点运行 kube-apiserver、kube-scheduler 和 kube-controller-manager 实例。 kube-apiserver 使用负载均衡器暴露给工作节点。
每个控制平面节点创建一个本地 etcd 成员(member)，这个 etcd 成员只与该节点的 kube-apiserver 通信。 这同样适用于本地 kube-controller-manager 和 kube-scheduler 实例。
这种拓扑将控制平面和 etcd 成员耦合在同一节点上。相对使用外部 etcd 集群， 设置起来更简单，而且更易于副本管理。
然而，堆叠集群存在耦合失败的风险。如果一个节点发生故障，则 etcd 成员和控制平面实例都将丢失， 并且冗余会受到影响。你可以通过添加更多控制平面节点来降低此风险。
因此，你应该为 HA 集群运行至少三个堆叠的控制平面节点。
┌──[root@vms100.liruilongs.github.io]-[~/ansible/kubescape]└─$ETCDCTL_API=3 etcdctl --endpoints https://127.0.0.1:2379 --cert="/etc/kubernetes/pki/etcd/server.crt" --key="/etc/kubernetes/pki/etcd/server.key" --cacert="/etc/kubernetes/pki/etcd/ca.crt" endpoint status --cluster -w table+-----------------------------+------------------+---------+---------+-----------+-----------+------------+| ENDPOINT | ID | VERSION | DB SIZE | IS LEADER | RAFT TERM | RAFT INDEX |+-----------------------------+------------------+---------+---------+-----------+-----------+------------+| https://192.168.26.100:2379 | ee392e5273e89e2 | 3.5.4 | 37 MB | false |100 | 3152364 || https://192.168.26.102:2379 | 11486647d7f3a17b | 3.5.4 | 36 MB | false |100 | 3152364 || https://192.168.26.101:2379 | e00e3877df8f76f4 | 3.5.4 | 36 MB | true |100 | 3152364 |+-----------------------------+------------------+---------+---------+-----------+-----------+------------+┌──[root@vms100.liruilongs.github.io]-[~/ansible/kubescape]└─$当使用 kubeadm init 和 kubeadm join --control-plane 时， 在控制平面节点上会自动创建本地 etcd 成员。ETCD 一点要定期备份
部署中涉及太多文件，篇幅问题，文件无法展示，以上传 git 仓库，获取 git 仓库地址方式，关注 公总好 山河已无恙，回复 k8s-ha-deploy 即可获得地址。
免密配置
这里假设我们拿到的是安装了系统的新机。这里我的机器，通过 克隆的的生成的 虚机，所以配置了 YUM 源，如果没有，需要先配置一台虚机，剩下的通过 Ansible 的批量配置下 YUM ，所以对应 YUM 的配置这个教程没有涉及，YUM 主要配置为 阿里云的 就可以。配置文件小伙伴可以通过 git 获取。
在安装 Ansible 之前，方便操作，我们需要批量配置下免密，这个使用 expect 的方式实现，用直接写好了 脚本直接执行，需要配置的主机 单独列出读取。
安装 expect
┌──[root@vms100.liruilongs.github.io]-[~]└─$yum -y install expect列出部署主机
┌──[root@vms100.liruilongs.github.io]-[~]└─$cat host_list192.168.26.100192.168.26.101192.168.26.102192.168.26.103192.168.26.105192.168.26.106192.168.26.107192.168.26.108免密脚本，直接执行即可，如果主机清单文件名相同，直接读取即可
#!/bin/bash#@File : mianmi.sh#@Time : 2022/08/20 17:45:53#@Author : Li Ruilong#@Version : 1.0#@Desc : None#@Contact : 1224965096@qq.com/usr/bin/expect <<-EOFspawn ssh-keygenexpect "(/root/.ssh/id_rsa)" {send "\r"}expect {"(empty for no passphrase)" {send "\r"}"already" {send "y\r"}}expect {"again" {send "\r"}"(empty for no passphrase)" {send "\r"}}expect {"again" {send "\r"}"#" {send "\r"}}expect "#"expect eofEOFfor IP in $( cat host_list )doif [ -n IP ];then/usr/bin/expect <<-EOFspawn ssh-copy-id root@$IPexpect {"*yes/no*" { send "yes\r"}"*password*" { send "redhat\r" }}expect {"*password" { send "redhat\r"}"#" { send "\r"}}expect "#"expect eofEOFfidoneOK，配置完免密之后，需要安装 Ansible， 运维工具，一定要装一个，机器太多了很费人。
Ansible 安装
┌──[root@vms100.liruilongs.github.io]-[~]└─$yum -y install ansibleansible 配置
这里的主机清单，随便配置一个无效的，建议不要在配置文件中配置正确的主机清单，防止执行剧本到错误的主机节点。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$vim ansible.cfg┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat ansible.cfg[defaults]# 主机清单文件，就是要控制的主机列表inventory=inventory# 连接受管机器的远程的用户名remote_user=root# 角色目录roles_path=roles# 设置用户的su 提权[privilege_escalation]become=Truebecome_method=sudobecome_user=rootbecome_ask_pass=False拷贝前面的文件测试下网络情况
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat host_list192.168.26.100192.168.26.101192.168.26.102192.168.26.103192.168.26.105192.168.26.106192.168.26.107192.168.26.108┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible all -m ping -i host_list测试没问题，需要编写下 主机清单文件
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat host.yamlansible: children: ansible_master: hosts: 192.168.26.100: ansible_node: hosts: 192.168.26.[101:103]: 192.168.26.[105:108]:k8s: children: k8s_master: hosts: 192.168.26.[100:102]: k8s_node: hosts: 192.168.26.103: 192.168.26.[105:108]:检查清单文件，这里的分组就不多讲了。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible-inventory -i host.yaml --graph@all: |--@ansible: | |--@ansible_master: | | |--192.168.26.100 | |--@ansible_node: | | |--192.168.26.101 | | |--192.168.26.102 | | |--192.168.26.103 | | |--192.168.26.105 | | |--192.168.26.106 | | |--192.168.26.107 | | |--192.168.26.108 |--@k8s: | |--@k8s_master: | | |--192.168.26.100 | | |--192.168.26.101 | | |--192.168.26.102 | |--@k8s_node: | | |--192.168.26.103 | | |--192.168.26.105 | | |--192.168.26.106 | | |--192.168.26.107 | | |--192.168.26.108 |--@ungrouped:测试一下清单文件
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible all --list-hosts -i host.yaml hosts (8): 192.168.26.100 192.168.26.101 192.168.26.102 192.168.26.103 192.168.26.105 192.168.26.106 192.168.26.107 192.168.26.108为了使用 tab 键，这里我们建一些清单组的空文件。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$touch ansible_master ansible_node k8s_master k8s_node┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$lsansible.cfg ansible_node host_list_no_ansible k8s_master ps1_mod.yamlansible_master host_list host.yaml k8s_node┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$为了方便演示，看到当前的执行目录，配置下 PS1
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat ps1_mod.yaml---- name: all modify PS1 hosts: ansible_node tasks: - name: PS1 modify shell: echo 'PS1="\[\033[1;32m\]┌──[\[\033[1;34m\]\u@\H\[\033[1;32m\]]-[\[\033[0;1m\]\w\[\033[1;32m\]] \n\[\033[1;32m\]└─\[\033[1;34m\]\$\[\033[0m\]"' >> /root/.bashrc - name: PS1 disply shell: cat /root/.bashrc | grep PS1执行剧本并测试。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible-playbook ps1_mod.yaml -i host.yaml -vv k8s 安装前环境准备
检查机器 UUID 和 网卡MAC 地址
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible k8s -m shell -a "ip link | grep ether| awk '{print $2 }' " -i host.yaml┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible k8s -m shell -a "cat /sys/class/dmi/id/product_uuid " -i host.yaml剩下的 操作这个编写一个 k8s 部署环境的初始化的 剧本任务 k8s_init_deploy.yaml,涉及各项通过剧本任务引用实现
hosts 文件准备
没有 DNS ，如果有 DNS 服务器，则不需要
┌──[root@vms100.liruilongs.github.io]-[~/ansible/file]└─$cat hosts127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4::1 localhost localhost.localdomain localhost6 localhost6.localdomain6192.168.26.100 vms100.liruilongs.github.io vms100192.168.26.101 vms101.liruilongs.github.io vms101192.168.26.102 vms102.liruilongs.github.io vms102192.168.26.103 vms103.liruilongs.github.io vms103192.168.26.105 vms105.liruilongs.github.io vms105192.168.26.106 vms106.liruilongs.github.io vms106192.168.26.107 vms107.liruilongs.github.io vms107192.168.26.108 vms108.liruilongs.github.io vms108把 hosts 文件替换为所有机器的 hosts ，编写剧本任务
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_init_deploy.yaml---- name: copy "/etc/hosts" copy: src: ./file/hosts dest: /etc/hosts force: yes防火墙，交换分区，SElinux 设置
编写剧本任务，在之前的 初始化剧本补充。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_init_deploy.yaml---# 拷贝 hosts 文件,没有 DNS 需要- name: copy "/etc/hosts" copy: src: ./file/hosts dest: /etc/hosts force: yes# 关闭防火墙,这里设置为 trusted ,以后可能处理漏洞使用- name: firewalld setting trusted firewalld: zone: trusted permanent: yes state: enabled# 关闭 SELinux- name: Disable SELinux selinux: state: disabled# 禁用交换分区- name: Disable swapoff shell: /usr/sbin/swapoff -a# 删除 swap 配置- name: delete /etc/fstab shell: sed -i '/swap/d' /etc/fstab这里我们通过 任务引用的方式来执行，并且打一个 init 的标签。方便之后单独执行，引入到当前的 k8s 部署剧本 k8s_deploy.yaml
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_deploy.yaml---- name: k8s deploy hosts: k8s tasks: - name: init k8s include_tasks: file: k8s_init_deploy.yaml tags: - init_k8s k8s_deploy.yaml 为部署 k8s 所有操作的剧本。可以先执行下，完成前面的部署工作。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible-playbook k8s_deploy.yaml -i host.yaml -f 4有 8 个 机器，这里的 -f 4 为并行执行的意思。 -i host.yaml 指定主机清单
安装容器运行时 CRI docker
这里选择 Dokcer 作为 CRI ，但是Docker 本身没有实现 CRI，在 k8s 在 1.24 的移除了 docker 和 K8s 的桥梁 Dockershim ，所以不能直接使用，需要安装 cri-docker .
这里需要注意：
部署 cri-docker 要重载沙箱(pause)镜像，cri-dockerd 适配器能够接受指定用作 Pod 的基础容器的容器镜像(“pause 镜像”)作为命令行参数。 要使用的命令行参数是 --pod-infra-container-image。 如果不指定，会直接从 谷歌的镜像库拉取，即使在 kubeadm init 指定了 镜像库也不行(我部署是不行)，同时，如果部署 HA ，使用 keepalived + Haproxy 的方式，并且通过 静态 Pod 的方式部署，在 启动 kubelet 的时候 静态pod 也会直接从 谷歌的镜像库拉取。
对应 cgroup 驱动的使用，按照官方官网推荐的来。 如果系统通过 systemd 来引导，就用 systemd
CRI 安装的初始化工作
转发 IPv4 并让 iptables 看到桥接流量, 相关文件准备，需要注意
┌──[root@vms100.liruilongs.github.io]-[~/ansible/file/modules-load.d]└─$cat k8s.confoverlaybr_netfilter┌──[root@vms100.liruilongs.github.io]-[~/ansible/file/sysctl.d]└─$cat k8s.confnet.bridge.bridge-nf-call-iptables = 1net.bridge.bridge-nf-call-ip6tables = 1net.ipv4.ip_forward = 1docker 配置文件准备，这里的配置项 "data-root": "/docker/data", 为 docker 数据目录，最好找一个大一点的地方。如果修改的话，剧本的创建目录要同步修改。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat file/daemon.json{ "exec-opts": ["native.cgroupdriver=systemd"], "log-driver": "json-file", "data-root": "/docker/data", "log-opts": { "max-size": "100m" }, "storage-driver": "overlay2", "storage-opts": [ "overlay2.override_kernel_check=true" ], "experimental": false, "debug": false, "max-concurrent-downloads": 10, "registry-mirrors": ["https://2tefyfv7.mirror.aliyuncs.com","https://docker.mirrors.ustc.edu.cn" ]}当前需要复制的配置文件
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ls file/*file/daemon.json file/hostsfile/modules-load.d:k8s.conffile/sysctl.d:k8s.confcri-dockerd rpm 包下载，下载到 Anasible 控制节点，复制到其他的节点。
PS C:\Users\山河已无恙\Downloads> curl -o cri-dockerd-0.3.0-3.el7.x86_64.rpm https://github.com/Mirantis/cri-dockerd/releases/download/v0.3.0/cri-dockerd-0.3.0-3.el7.x86_64.rpmPS C:\Users\山河已无恙\Downloads> scp .\cri-dockerd-0.3.0-3.el7.x86_64.rpm root@192.168.26.100:/root/ansible/install_packageroot@192.168.26.100's password:cri-dockerd-0.3.0-3.el7.x86_64.rpm100% 9195KB 81.7MB/s 00:00PS C:\Users\山河已无恙\Downloads>存放位置
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cd install_package/;lscri-dockerd-0.3.0-3.el7.x86_64.rpmCRI docker, 安装 任务剧本
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_cri_deploy.yaml# ansible 2.9.27# #@File : k8s_cri_deploy.yaml# #@Time : 2023/01/19 22:32:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : 安装 CRI , 这里我们选择 Docker,需要安装 cri-docker# #@Contact : liruilonger@gmail.com---# 转发 IPv4 并让 iptables 看到桥接流量- name: Forwarding IPv4 and letting iptables see bridged traffic 1 copy: src: ./file/modules-load.d/k8s.conf dest: /etc/modules-load.d/k8s.conf- name: Forwarding IPv4 and letting iptables see bridged traffic 2 shell: modprobe overlay && modprobe br_netfilter# 永久设置所需的 sysctl 参数- name: sysctl params required by setup, params persist across reboots copy: src: ./file/sysctl.d/k8s.conf dest: /etc/sysctl.d/k8s.conf# 刷新内核参数- name: Apply sysctl params without reboot shell: sysctl --system# 安装 docker- name: install docker-ce docker-ce-cli containerd.io yum: name: - docker-ce - docker-ce-cli - containerd.io state: present# 创建 dockers 数据目录- name: create docker data dir file: path: /docker/data state: directory# 创建 docker 配置目录- name: create docker data dir file: path: /etc/docker state: directory# 修改 dockers 配置- name: modify docker config copy: src: ./file/daemon.json dest: /etc/docker/daemon.json# 复制 cri-docker rpm 安装包- name: copy install cri-docker rpm # https://github.com/Mirantis/cri-dockerd/releases/download/v0.3.0/cri-dockerd-0.3.0-3.el7.x86_64.rpm copy: src: ./install_package/cri-dockerd-0.3.0-3.el7.x86_64.rpm dest: /tmp/cri-dockerd-0.3.0-3.el7.x86_64.rpm# 安装 cri-docker- name: install cri-docker yum: name: /tmp/cri-dockerd-0.3.0-3.el7.x86_64.rpm state: present# 修改 cri-docker service 文件，否则 沙箱会从谷歌的仓库拉去- name: modify cri-dockerd service file copy: src: ./file/cri-docker.service dest: /usr/lib/systemd/system/cri-docker.service# 配置开机自启,启动 docker- name: start docker, setting enable service: name: docker state: restarted enabled: yes# 配置开机自启,启动 cri-docker- name: start cri-docker, setting enable service: name: cri-docker state: restarted enabled: yes# 配置开机自启,启动 cri-docker.socket- name: start cri-docker,socket setting enable service: name: cri-docker.socket enabled: yes# 配置校验- name: check init cri include_tasks: file: k8s_cri_deploy_check.yaml tags: cri_check┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$安装完的校验单独放到了 k8s_cri_deploy_check.yaml 位置，也是通过剧本引用的方式，并且打了 cri_check 标签。这个如果希望看到检查数据，需要执行剧本时加上 -vv,这里只是写了几个基本的校验，还可以对其他的做校验
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_cri_deploy_check.yaml# ansible 2.9.27# #@File : k8s_cri_deploy_check.yaml# #@Time : 2023/01/19 23:02:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : cri(docker) 配置检查剧本任务# #@Contact : liruilonger@gmail.com---- name: check br_netfilter, overlay shell: (lsmod | grep br_netfilter ;lsmod | grep overlay)- name: charck ipv4 sysctl param shell: sysctl net.bridge.bridge-nf-call-iptables net.bridge.bridge-nf-call-ip6tables net.ipv4.ip_forward- name: chrck docker config shell: docker info当前的 k8s 安装剧本，引入了 k8s_cri_deploy.yaml 的安装，并且 打标签 cri。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_deploy.yaml# ansible 2.9.27# #@File : k8s_deploy.yaml# #@Time : 2023/01/19 23:02:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : k8s 安装剧本# #@Contact : liruilonger@gmail.com---- name: k8s deploy 1 hosts: k8s tasks: # 初始化 K8s 安装环境 - name: init k8s include_tasks: file: k8s_init_deploy.yaml tags: - init_k8s # 安装 CRI docker ,cri-docker - name: CRI deploy (docker,cri-docker) include_tasks: file: k8s_cri_deploy.yaml tags: cri可以执行测试下
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible-playbook k8s_deploy.yaml -i host.yaml -f 4安装 kubeadm、kubelet 和 kubectl
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_kubeadm_kubelet_kubectl_deploy.yaml# ansible 2.9.27# #@File : k8s_kubeadm_kubelet_kubectl_deploy.yaml# #@Time : 2023/01/19 23:30:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : 安装 kubeadm、kubelet 和 kubectl# #@Contact : liruilonger@gmail.com---# 安装 kubeadm、kubelet 和 kubectl- name: install kubeadm、kubelet and kubectl yum: name: - kubelet-1.25.1-0 - kubeadm-1.25.1-0 - kubectl-1.25.1-0 state: present# 启动 kubelet 并设置开启自启- name: start kubelet,setting enable service: name: kubelet state: started enabled: yes┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$添加的部署脚本
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_deploy.yaml# ansible 2.9.27# #@File : k8s_deploy.yaml# #@Time : 2023/01/19 23:02:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : k8s 安装剧本# #@Contact : liruilonger@gmail.com---- name: k8s deploy 1 hosts: k8s tasks: # 初始化 K8s 安装环境 - name: init k8s include_tasks: file: k8s_init_deploy.yaml tags: - init_k8s # 安装 CRI docker ,cri-docker - name: CRI deploy (docker,cri-docker) include_tasks: file: k8s_cri_deploy.yaml tags: cri # 安装 kubeadm,kubelet,kubectl - name: install kubeadm,kubelet,kubectl include_tasks: file: k8s_kubeadm_kubelet_kubectl_deploy.yaml tags: install_k8s┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$HA 涉及的 keepalived、HAproxy 静态 Pod 准备
HA 涉及的 keepalived、HAproxy 通过 静态Pod 的方式运行，当前集群设置三个master 节点，kubelet 和 api-service 交互通过 keepalived 提供的 VIP 访问。然后由 HAprxy 把VIP地址反向代理到各 master 的 api-service 服务
涉及的到文件太多，这个不做展示
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$tree -h file/file/├── [ 25] haproxy│ └── [1.8K] haproxy.cfg├── [ 55] keepalived│ ├── [ 370] check_apiserver.sh│ └── [ 521] keepalived.conf└── [ 49] manifests ├── [ 543] haproxy.yaml └── [ 676] keepalived.yaml对应的 任务剧本 manifests_keepalived_haproxy.yaml
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat manifests_keepalived_haproxy.yaml# ansible 2.9.27# # #@File : manifests_keepalived_haproxy.yaml# # #@Time : 2023/01/19 23:02:47# # #@Author : Li Ruilong# # #@Version : 1.0# # #@Desc : HA 静态 pod 相关 配置文件 YAML 文件准备# # #@Contact : liruilonger@gmail.com---# 创建 静态 pod 目录- name: create manifests dir file: path: /etc/kubernetes/manifests/ state: directory force: true# 复制 keepalived haproxy 对应的 静态 pod yaml 文件- name: copy manifests pod, haproxy and keepalived copy: src: ./file/manifests/keepalived.yaml dest: /etc/kubernetes/manifests/keepalived.yaml- name: copy manifests pod, haproxy and keepalived copy: src: ./file/manifests/haproxy.yaml dest: /etc/kubernetes/manifests/haproxy.yaml# 创建 haproxy 配置文件 目录- name: create haproxy dir file: path: /etc/haproxy state: directory force: true- name: copy /etc/haproxy/haproxy.cfg copy: src: ./file/haproxy/haproxy.cfg dest: /etc/haproxy/haproxy.cfg# 创建 keepalived 配置文件 目录- name: create keepalived dir file: path: /etc/keepalived state: directory force: true- name: copy /etc/keepalived/keepalived.conf copy: src: ./file/keepalived/keepalived.conf dest: /etc/keepalived/keepalived.conf- name: copy /etc/keepalived/check_apiserver.sh copy: src: ./file/keepalived/check_apiserver.sh dest: /etc/keepalived/check_apiserver.sh┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$涉及到 HAproxy 和 keepalived 的 配置文件以及静态pod yaml文件的定义，篇幅原因，这里不做展示，可以参考官方文档给出的，如果实在不想找， 也可以通过 git 地址获取。
这个时候的部署剧本
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$cat k8s_deploy.yaml# ansible 2.9.27# #@File : k8s_deploy.yaml# #@Time : 2023/01/19 23:02:47# #@Author : Li Ruilong# #@Version : 1.0# #@Desc : k8s 安装剧本# #@Contact : liruilonger@gmail.com---- name: k8s deploy 1 hosts: k8s tasks: # 初始化 K8s 安装环境 - name: init k8s include_tasks: file: k8s_init_deploy.yaml tags: - init_k8s # 安装 CRI docker ,cri-docker - name: CRI deploy (docker,cri-docker) include_tasks: file: k8s_cri_deploy.yaml tags: cri # 安装 kubeadm,kubelet,kubectl - name: install kubeadm,kubelet,kubectl include_tasks: file: k8s_kubeadm_kubelet_kubectl_deploy.yaml tags: install_k8s- name: k8s deploy 2 hosts: k8s_master tasks: - name: manifests_keepalived_haproxy.yaml include_tasks: file: manifests_keepalived_haproxy.yaml tags: install_k8s初始化集群添加的控制节点，工作节点
先在一个 maste 节点执行，剩下的 master 节点通过添加的加入
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$kubeadm init --image-repository "registry.aliyuncs.com/google_containers" --control-plane-endpoint "192.168.26.99:30033" --upload-certs --kubernetes-version=v1.25.1 --pod-network-cidr=10.244.0.0/16 --cri-socket /var/run/cri-dockerd.sockW0121 02:49:21.251663 106843 initconfiguration.go:119] Usage of CRI endpoints without URL scheme is deprecated and can cause kubelet errors in the future. Automatically prepending scheme "unix" to the "criSocket" with value "/var/run/cri-dockerd.sock". Please update your configuration![init] Using Kubernetes version: v1.25.1[preflight] Running pre-flight checks [WARNING Firewalld]: firewalld is active, please ensure ports [6443 10250] are open or your cluster may not function correctly[preflight] Pulling images required for setting up a Kubernetes cluster......................[addons] Applied essential addon: CoreDNSW0121 02:49:31.282997 106843 endpoint.go:57] [endpoint] WARNING: port specified in controlPlaneEndpoint overrides bindPort in the controlplane address[addons] Applied essential addon: kube-proxyYour Kubernetes control-plane has initialized successfully!To start using your cluster, you need to run the following as a regular user: mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/configAlternatively, if you are the root user, you can run: export KUBECONFIG=/etc/kubernetes/admin.confYou should now deploy a pod network to the cluster.Run "kubectl apply -f [podnetwork].yaml" with one of the options listed at: https://kubernetes.io/docs/concepts/cluster-administration/addons/You can now join any number of the control-plane node running the following command on each as root: kubeadm join 192.168.26.99:30033 --token hdse4x.7d8vdil1s7tdp8dj \ --discovery-token-ca-cert-hash sha256:11bf0456c84c2a176eb380a98844eed80a296cdb6cc557c863991c63de594108 \ --control-plane --certificate-key 61b1a22aa5dbbd93660709aa343452d80ec98bf2efef6915b3a36ae97a8eefedPlease note that the certificate-key gives access to cluster sensitive data, keep it secret!As a safeguard, uploaded-certs will be deleted in two hours; If necessary, you can use"kubeadm init phase upload-certs --upload-certs" to reload certs afterward.Then you can join any number of worker nodes by running the following on each as root:kubeadm join 192.168.26.99:30033 --token hdse4x.7d8vdil1s7tdp8dj \ --discovery-token-ca-cert-hash sha256:11bf0456c84c2a176eb380a98844eed80a296cdb6cc557c863991c63de594108第一个控制节点 准备 kubectl 客户端
mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/configCNI 插件准备
k8s 版本 和 CNI 的版本关系，以及安装相关可以通过下的地址查看。
https://projectcalico.docs.tigera.io/archive/v3.24/getting-started/kubernetes/self-managed-onprem/onpremises
https://projectcalico.docs.tigera.io/archive/v3.24/getting-started/kubernetes/requirements
这里需要注意的是:
好像 bpf 文件系统，需要高版本的 内核才支撑。我最开始的内核版本为 3.10 的版本，所有不支持，需要把 bpf 的挂载删掉。
我本地测试发现 calico v3.20 到 v3.25 都是需要的做处理的。 涉及的镜像最好导入进去，网络情况不同，拉取很费事。
┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$cat calico.v3.23.yaml | grep -A 3 -e bpffs$ - name: bpffsmountPath: /sys/fs/bpf - name: cni-log-dirmountPath: /var/log/calico/cni-- - name: bpffs hostPath: path: /sys/fs/bpf type: Directory需要删除的部分
。。。。。 - name: bpffsmountPath: /sys/fs/bpf。。。。。 - name: bpffs hostPath: path: /sys/fs/bpf type: Directory下载 YAML 文件部署
┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$wget --no-check-certificate https://docs.projectcalico.org/manifests/calico.yaml--2023-01-21 02:56:01-- https://docs.projectcalico.org/manifests/calico.yaml正在解析主机 docs.projectcalico.org (docs.projectcalico.org)... 18.139.194.139, 52.74.166.77, 2406:da18:880:3801::c8, ...正在连接 docs.projectcalico.org (docs.projectcalico.org)|18.139.194.139|:443... 已连接。警告: 无法验证 docs.projectcalico.org 的由 “/C=US/O=Let's Encrypt/CN=R3” 颁发的证书: 颁发的证书已经过期。已发出 HTTP 请求，正在等待回应... 200 OK长度：238089 (233K) [text/yaml]正在保存至: “calico.yaml”100%[=======================================================================================================================================================================>] 238,089 592KB/s 用时 0.4s2023-01-21 02:56:02 (592 KB/s) - 已保存 “calico.yaml” [238089/238089])┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$vim calico.yaml修改 CALICO_IPV4POOL_CIDR 为之前指定的 地址
┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$cat calico.yaml | grep -C 3 IPV4POOL_CIDR # The default IPv4 pool to create on startup if none exists. Pod IPs will be # chosen from this range. Changing this value after installation will have # no effect. This should fall within `--cluster-cidr`. - name: CALICO_IPV4POOL_CIDRvalue: "10.244.0.0/16" # Disable file logging so `kubectl logs` works. - name: CALICO_DISABLE_FILE_LOGGING┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$kubectl apply -f ./calico/calico.yaml┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$source <(kubectl completion bash) >> /etc/profile其他控制节点添加
这个只有3个控制节点，所以通过 命令行单独处理，多的话可以使用 ansible
┌──[root@vms101.liruilongs.github.io]-[~]└─$kubeadm join 192.168.26.99:30033 --token hdse4x.7d8vdil1s7tdp8dj --discovery-token-ca-cert-hash sha256:11bf0456c84c2a176eb380a98844eed80a296cdb6cc557c863991c63de594108 --control-plane --certificate-key 61b1a22aa5dbbd93660709aa343452d80ec98bf2efef6915b3a36ae97a8eefed --cri-socket /var/run/cri-dockerd.sock......┌──[root@vms102.liruilongs.github.io]-[~]└─$kubeadm join 192.168.26.99:30033 --token hdse4x.7d8vdil1s7tdp8dj --discovery-token-ca-cert-hash sha256:11bf0456c84c2a176eb380a98844eed80a296cdb6cc557c863991c63de594108 --control-plane --certificate-key 61b1a22aa5dbbd93660709aa343452d80ec98bf2efef6915b3a36ae97a8eefed --cri-socket /var/run/cri-dockerd.sock拷贝 kubeconfig 文件并 配置 命令补全。每个 master 节点执行。
mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/config source <(kubectl completion bash) >> /etc/profile┌──[root@vms101.liruilongs.github.io]-[~]└─$工作节点添加
工作节点通过 ansible 并行批量添加。
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ansible k8s_node -m shell -a "kubeadm join 192.168.26.99:30033 --token hdse4x.7d8vdil1s7tdp8dj --discovery-token-ca-cert-hash sha256:11bf0456c84c2a176eb380a98844eed80a296cdb6cc557c863991c63de594108 --cri-socket /var/run/cri-dockerd.sock " -i host.yaml查看节点状态
┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$kubectl get nodesNAME STATUS ROLES AGE VERSIONvms100.liruilongs.github.io Ready control-plane 17h v1.25.1vms101.liruilongs.github.io Ready control-plane 15h v1.25.1vms102.liruilongs.github.io Ready control-plane 15h v1.25.1vms103.liruilongs.github.io Ready <none> 15h v1.25.1vms105.liruilongs.github.io Ready <none> 15h v1.25.1vms106.liruilongs.github.io Ready <none> 15h v1.25.1vms107.liruilongs.github.io Ready <none> 15h v1.25.1vms108.liruilongs.github.io Ready <none> 15h v1.25.1┌──[root@vms100.liruilongs.github.io]-[~/ansible/calico]└─$到这里集群就算是安装完成，但是对于生产环境，我们需要安装一个常用的插件，集群备份，监控工具，Ingress 控制器等。这部分属于可选项。
部署后的可选操作
安装一些常用插件和工具
一些常用的工具和 kubelct 插件，这里我直接从旧的集群里拷贝过来。安装相对简单，没有网络的可以，下载二进制包，然后配置成 kubectl 插件。有的话可以先下载 krew ，通过 krew 下载其他的插件
┌──[root@vms100.liruilongs.github.io]-[/usr/local/bin]└─$tree -h.├── [ 45M] helm├── [ 15M] helmify├── [9.2M] kubectl-ketall├── [ 11M] kubectl-krew├── [ 44M] kubectl-kubepug├── [8.8M] kubectl-rakkess├── [ 18M] kubectl-score├── [ 57M] kubectl-spy├── [ 31M] kubectl-tree├── [ 57M] kubectl-virt└── [ 14M] kustomize0 directories, 11 files工具介绍：
- helm: HELM chart 包管理器
- kustomize： YAML 文件整合管理工具，用于管理整合生成 资源 YAML文件
- helmify： YAML 文件转 HELM chart 包工具，可以把 YAML 文件转化为 charts 包
- kubectl-ketall： 查看所有集群资源的 kubelct 插件工具
- kubectl-krew： kubelet 插件管理工具，用于自动的安装升级 kubectl 插件。
- kubectl-kubepug： 集群API资源版本 查看，用于升级检查
- kubectl-rakkess： 集群 RBAC 权限查看工具，用于查看整个集群授权
- kubectl-score： API资源定义建议工具，用于给出一些 API 资源的优化建议。
- kubectl-spy： 集群 API 资源动态监控工具，可以看到具体的YAML 字段变动。
- kubectl-tree： 集群 API 资源 层级关系，用于展示 API 资源的 树状关系
- kubectl-virt: 集群虚机环境管理工具，用于管理接入集群中的虚拟机。
安装一些监控工具
cadvisor 安装
SPS 1.25不被支持，需要提前去掉，或者看看下官网的通过 kustomize 修改后安装
┌──[root@vms100.liruilongs.github.io]-[~/ansible/cadvisor]└─$kubectl apply -f cadvisor.yamlnamespace/cadvisor createdserviceaccount/cadvisor createdclusterrole.rbac.authorization.k8s.io/cadvisor createdclusterrolebinding.rbac.authorization.k8s.io/cadvisor createddaemonset.apps/cadvisor created安装 metrics-server 监控工具
┌──[root@vms100.liruilongs.github.io]-[~/ansible/cadvisor]└─$kubectl apply -f metrics-server.yamlserviceaccount/metrics-server createdclusterrole.rbac.authorization.k8s.io/system:aggregated-metrics-reader createdclusterrole.rbac.authorization.k8s.io/system:metrics-server createdrolebinding.rbac.authorization.k8s.io/metrics-server-auth-reader createdclusterrolebinding.rbac.authorization.k8s.io/metrics-server:system:auth-delegator createdclusterrolebinding.rbac.authorization.k8s.io/system:metrics-server createdservice/metrics-server createddeployment.apps/metrics-server createdapiservice.apiregistration.k8s.io/v1beta1.metrics.k8s.io created安装 ingress-nginx 用于 Ingress
为了让 Ingress 资源工作，集群必须有一个正在运行的 Ingress 控制器。
与作为 kube-controller-manager 可执行文件的一部分运行的其他类型的控制器不同， Ingress 控制器不是随集群自动启动的。 基于此页面，你可选择最适合你的集群的 ingress 控制器实现。
Kubernetes 作为一个项目，目前支持和维护 AWS、 GCE 和 Nginx Ingress 控制器
┌──[root@vms100.liruilongs.github.io]-[~/ansible/ingress_nginx]└─$helm upgrade --install ingress-nginx ingress-nginx --repo https://kubernetes.github.io/ingress-nginx --namespace ingress-nginx --create-namespace如果没有网，可以把 yaml 文件和 对应的 镜像导入进去。安装完成有个 Service 类型是 LB ,所有我们还的安装一个软 LB
┌──[root@vms100.liruilongs.github.io]-[~/ansible/ingress_nginx]└─$kubectl apply -f deploy.yamlnamespace/ingress-nginx createdserviceaccount/ingress-nginx createdserviceaccount/ingress-nginx-admission createdrole.rbac.authorization.k8s.io/ingress-nginx createdrole.rbac.authorization.k8s.io/ingress-nginx-admission createdclusterrole.rbac.authorization.k8s.io/ingress-nginx createdclusterrole.rbac.authorization.k8s.io/ingress-nginx-admission createdrolebinding.rbac.authorization.k8s.io/ingress-nginx createdrolebinding.rbac.authorization.k8s.io/ingress-nginx-admission createdclusterrolebinding.rbac.authorization.k8s.io/ingress-nginx createdclusterrolebinding.rbac.authorization.k8s.io/ingress-nginx-admission createdconfigmap/ingress-nginx-controller createdservice/ingress-nginx-controller createdservice/ingress-nginx-controller-admission createddeployment.apps/ingress-nginx-controller createdjob.batch/ingress-nginx-admission-create createdjob.batch/ingress-nginx-admission-patch createdingressclass.networking.k8s.io/nginx createdvalidatingwebhookconfiguration.admissionregistration.k8s.io/ingress-nginx-admission created安装 Metallb 用于 LoadBalancer
Metallb 实现 LoadBalancer
Metallb可以通过k8s原生的方式提供LB类型的Service支持
kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.13.7/config/manifests/metallb-native.yaml┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$kubectl apply -f metallb-native.yamlnamespace/metallb-system createdcustomresourcedefinition.apiextensions.k8s.io/addresspools.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/bfdprofiles.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/bgpadvertisements.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/bgppeers.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/communities.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/ipaddresspools.metallb.io createdcustomresourcedefinition.apiextensions.k8s.io/l2advertisements.metallb.io createdserviceaccount/controller createdserviceaccount/speaker createdrole.rbac.authorization.k8s.io/controller createdrole.rbac.authorization.k8s.io/pod-lister createdclusterrole.rbac.authorization.k8s.io/metallb-system:controller createdclusterrole.rbac.authorization.k8s.io/metallb-system:speaker createdrolebinding.rbac.authorization.k8s.io/controller createdrolebinding.rbac.authorization.k8s.io/pod-lister createdclusterrolebinding.rbac.authorization.k8s.io/metallb-system:controller createdclusterrolebinding.rbac.authorization.k8s.io/metallb-system:speaker createdsecret/webhook-server-cert createdservice/webhook-service createddeployment.apps/controller createddaemonset.apps/speaker createdvalidatingwebhookconfiguration.admissionregistration.k8s.io/metallb-webhook-configuration created创建 IP池
┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$kubectl apply -f pool.yamlipaddresspool.metallb.io/first-pool created┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$cat pool.yamlapiVersion: metallb.io/v1beta1kind: IPAddressPoolmetadata: name: first-pool namespace: metallb-systemspec: addresses: - 192.168.26.220-192.168.26.249┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$cat l2a.yamlapiVersion: metallb.io/v1beta1kind: L2Advertisementmetadata: name: example namespace: metallb-system┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$kubectl apply -f l2a.yamll2advertisement.metallb.io/example unchanged┌──[root@vms100.liruilongs.github.io]-[~/ansible/metallb]└─$安装一个本地存储的 SC
我们需要安装一个 SC 分配器，以后可能会用到，这个也可以以后安装 : https://github.com/rancher/local-path-provisioner
┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$kubectl apply -f local-path-storage.yamlnamespace/local-path-storage unchangedserviceaccount/local-path-provisioner-service-account unchangedclusterrole.rbac.authorization.k8s.io/local-path-provisioner-role unchangedclusterrolebinding.rbac.authorization.k8s.io/local-path-provisioner-bind unchangeddeployment.apps/local-path-provisioner unchangedstorageclass.storage.k8s.io/local-path unchangedconfigmap/local-path-config unchanged安装 prometheus
需要指标监控，所有需要普罗米修斯
kube-prometheus-stack-30.0.1
https://github.com/prometheus-community/helm-charts/releases/download/kube-prometheus-stack-30.0.1/kube-prometheus-stack-30.0.1.tgz
镜像拉不了的问题，直接替换不好找，这里把 charts 包下载下来，然后通过 helm template 转化为具体的 yaml 文件。替换对应的镜像。但是这样还一个问题，一些 CRD 不会预先安装，尤其是多 master 的情况，这里你可以多试几次，说不定就可以了，github上有人提了，貌似没有很好的解决方案，我的解决办法是先用 helm 安装，然后卸载，卸载的时候不会卸载 crd，然后在运行 生成的 yaml 文件。
┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$cd kube-prometheus-stack┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm/kube-prometheus-stack]└─$lsChart.lock charts Chart.yaml CONTRIBUTING.md crds README.md templates values.yaml┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm/kube-prometheus-stack]└─$helm install kube-prometheus-stack .┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm/kube-prometheus-stack]└─$helm template . > ../kube-prometheus-stack.yaml执行应用
┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$kubectl apply -f kube-prometheus-stack.yaml执行完需要修改svc 为 NodePort 当然如果有 Ingress 控制器，或者 LB ，可以配置其他类型。
┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$kubectl get svcNAME TYPE CLUSTER-IPEXTERNAL-IP PORT(S) AGEalertmanager-operatedClusterIP None <none> 9093/TCP,9094/TCP,9094/UDP 27mkubernetes ClusterIP 10.96.0.1 <none> 443/TCP 40hprometheus-operated ClusterIP None <none> 9090/TCP27mrelease-name-grafana NodePort 10.111.188.209 <none> 80:30203/TCP 30mrelease-name-kube-promethe-alertmanager ClusterIP 10.97.17.175 <none> 9093/TCP30mrelease-name-kube-promethe-operatorClusterIP 10.107.60.174 <none> 443/TCP 30mrelease-name-kube-promethe-prometheus ClusterIP 10.108.163.61 <none> 9090/TCP30mrelease-name-kube-state-metrics ClusterIP 10.102.37.208 <none> 8080/TCP30mrelease-name-prometheus-node-exporter ClusterIP 10.100.5.155 <none> 9100/TCP30m登录用户名和密码获取
┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$kubectl get secrets release-name-grafana -o jsonpath='{.data.admin-user}' | base64 -dadmin┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$kubectl get secrets release-name-grafana -o jsonpath='{.data.admin-password}' | base64 -dprom-operator┌──[root@vms100.liruilongs.github.io]-[~/ansible/helm]└─$ETCD 快照文件定时备份
生产环境的 ETCD 一定要做备份，要不出了问题只能跑路了…
service 服务编写
这里我们写了一个脚本，通过 systemd.service 的方式运行。存放位置见注释
┌──[root@vms81.liruilongs.github.io]-[~/back]└─$systemctl cat etcd-backup# /usr/lib/systemd/system/etcd-backup.service[Unit]Description= "ETCD 备份"After=network-online.target[Service]Type=oneshotEnvironment=ETCDCTL_API=3ExecStart=/usr/bin/bash /usr/lib/systemd/system/etcd_back.sh[Install]WantedBy=multi-user.target定时任务编写
定时备份通过 systemd.timer 的方式实现
┌──[root@vms81.liruilongs.github.io]-[~/back]└─$systemctl cat etcd-backup.timer# /usr/lib/systemd/system/etcd-backup.timer[Unit]Description="每天备份一次 ETCD"[Timer]OnBootSec=3sOnCalendar=*-*-* 00:00:00Unit=etcd-backup.service[Install]WantedBy=multi-user.target备份脚本编写
具体的 ETCD 快照备份脚本
┌──[root@vms81.liruilongs.github.io]-[~/back]└─$cat /usr/lib/systemd/system/etcd_back.sh#!/bin/bash#@File : erct_break.sh#@Time : 2023/01/27 23:00:27#@Author : Li Ruilong#@Version : 1.0#@Desc : ETCD 备份#@Contact : 1224965096@qq.comif [ ! -d /root/back/ ];then mkdir -p /root/back/fiSTR_DATE=$(date +%Y%m%d%H%M)ETCDCTL_API=3 etcdctl \--endpoints="https://127.0.0.1:2379" \--cert="/etc/kubernetes/pki/etcd/server.crt" \--key="/etc/kubernetes/pki/etcd/server.key" \--cacert="/etc/kubernetes/pki/etcd/ca.crt" \snapshot save /root/back/snap-${STR_DATE}.dbETCDCTL_API=3 etcdctl --write-out=table snapshot status /root/back/snap-${STR_DATE}.db运行方式
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$systemctl enable etcd-backup.service --now┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$systemctl enable etcd-backup.timer --now查看备份情况
┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$ls -lh /root/back/总用量 311M-rw-r--r-- 1 root root 27M 1月 28 00:17 snap-202301280017.db-rw-r--r-- 1 root root 27M 1月 29 00:00 snap-202301290000.db-rw-r--r-- 1 root root 27M 2月 1 21:43 snap-202302012143.db-rw-r--r-- 1 root root 27M 2月 2 00:00 snap-202302020000.db-rw-r--r-- 1 root root 29M 2月 3 00:00 snap-202302030000.db-rw-r--r-- 1 root root 29M 2月 4 00:00 snap-202302040000.db-rw-r--r-- 1 root root 36M 2月 5 00:00 snap-202302050000.db┌──[root@vms100.liruilongs.github.io]-[~/ansible]└─$安装 Velero 集群备份
Velero 用于集群 安全备份和恢复、执行灾难恢复以及迁移 Kubernetes 集群资源和持久卷。
客户端
┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero]└─$wget --no-check-certificate https://github.com/vmware-tanzu/velero/releases/download/v1.10.1-rc.1/velero-v1.10.1-rc.1-linux-amd64.tar.gz┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero]└─$lsvelero-v1.10.1-rc.1-linux-amd64.tar.gz┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero]└─$tar -zxvf velero-v1.10.1-rc.1-linux-amd64.tar.gz┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero]└─$cd velero-v1.10.1-rc.1-linux-amd64/┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$cp velero /usr/local/bin/┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$velero versionClient: Version: v1.10.1-rc.1 Git commit: e4d2a83917cd848e5f4e6ebc445fd3d262de10fa<error getting server version: no matches for kind "ServerStatusRequest" in version "velero.io/v1">服务端安装
配置登录相关的帐密文件
┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$vim credentials-velero┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$cat credentials-velero[default]aws_access_key_id = minioaws_secret_access_key = minio123启动服务器和本地存储服务。在 Velero 目录中，在上面的客户端的安装包里，解压出来就可以看到
修改下 yaml 文件，这个YAM 文件在上面下载的 安装包里。
┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$cat examples/minio/00-minio-deployment.yaml# Copyright 2017 the Velero contributors.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with the License.# You may obtain a copy of the License at## http://www.apache.org/licenses/LICENSE-2.0## Unless required by applicable law or agreed to in writing, software# distributed under the License is distributed on an "AS IS" BASIS,# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.# See the License for the specific language governing permissions and# limitations under the License.---apiVersion: v1kind: Namespacemetadata: name: velero---apiVersion: apps/v1kind: Deploymentmetadata: namespace: velero name: minio labels: component: miniospec: strategy: type: Recreate selector: matchLabels: component: minio template: metadata: labels: component: minio spec: volumes: - name: storage emptyDir: {} - name: config emptyDir: {} containers: - name: minio image: quay.io/minio/minio:latest imagePullPolicy: IfNotPresent args: - server - /storage - --console-address=:9090 - --config-dir=/config env: - name: MINIO_ROOT_USER value: "minio" - name: MINIO_ROOT_PASSWORD value: "minio123" ports: - containerPort: 9000 - containerPort: 9090 volumeMounts: - name: storage mountPath: "/storage" - name: config mountPath: "/config"---apiVersion: v1kind: Servicemetadata: namespace: velero name: minio labels: component: miniospec: # ClusterIP is recommended for production environments. # Change to NodePort if needed per documentation, # but only if you run Minio in a test/trial environment, for example with Minikube. type: NodePort ports: - port: 9000 name: api targetPort: 9000 protocol: TCP - port: 9099 name: console targetPort: 9090 protocol: TCP selector: component: minio---apiVersion: batch/v1kind: Jobmetadata: namespace: velero name: minio-setup labels: component: miniospec: template: metadata: name: minio-setup spec: restartPolicy: OnFailure volumes: - name: config emptyDir: {} containers: - name: mc image: minio/mc:latest imagePullPolicy: IfNotPresent command: - /bin/sh - -c - "mc --config-dir=/config config host add velero http://minio:9000 minio minio123 && mc --config-dir=/config mb -p velero/velero" volumeMounts: - name: config mountPath: "/config"不建议使用 empty dir 的方式。如果集群挂掉，很有可能无法重启 pod。以及对应的 容器。建议使用单独的容器部署挂载目录，
直接应用上面的 YAML 文件，访问查看。这里安装一个 minio 并且通过 job 添加了一个 桶，用于存放 备份后的数据。
通过命令行工具安装 velero
velero install \ --provider aws \ --plugins velero/velero-plugin-for-aws:v1.2.1 \ --bucket velero \ --secret-file ./credentials-velero \ --use-volume-snapshots=false \ --backup-location-config region=minio,s3ForcePathStyle="true",s3Url=http://minio.velero.svc:9000- bucket：你在minio中创建的bucketname
- backup-location-config: 把xxx.xxx.xxx.xxx改成你minio服务器的ip地址。
也可以导出 YAML 文件在应用，确认没问题，或者私库需要替换相关镜像
┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$velero install \ --provider aws \ --plugins velero/velero-plugin-for-aws:v1.2.1 \ --bucket velero \ --secret-file ./credentials-velero \ --use-volume-snapshots=false \ --backup-location-config region=minio,s3ForcePathStyle="true",s3Url=http://minio.velero.svc:9000 --dry-run -o yaml > velero_deploy.yaml┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$kubectl apply -f velero_deploy.yamlCustomResourceDefinition/backuprepositories.velero.io: attempting to create resourceCustomResourceDefinition/backuprepositories.velero.io: attempting to create resource client..........BackupStorageLocation/default: attempting to create resourceBackupStorageLocation/default: attempting to create resource clientBackupStorageLocation/default: createdDeployment/velero: attempting to create resourceDeployment/velero: attempting to create resource clientDeployment/velero: createdVelero is installed! ⛵ Use 'kubectl logs deployment/velero -n velero' to view the status.┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$部署完成的 job 会自动新建
备份
┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$velero backup create velero-demoBackup request "velero-demo" submitted successfully.Run `velero backup describe velero-demo` or `velero backup logs velero-demo` for more details.┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$velero get backup velero-demoNAME STATUSERRORS WARNINGS CREATED EXPIRES STORAGE LOCATION SELECTORvelero-demo InProgress 0 0 2023-01-28 22:18:45 +0800 CST 29ddefault <none>┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$┌──[root@vms100.liruilongs.github.io]-[~/ansible/velero/velero-v1.10.1-rc.1-linux-amd64]└─$velero get backup velero-demoNAME STATUS ERRORS WARNINGS CREATED EXPIRES STORAGE LOCATION SELECTORvelero-demo Completed 0 0 2023-01-28 22:18:45 +0800 CST 29ddefault <none>
篇幅原因，这块具体 velero 的 备份恢复,定时备份等不多说明。
关于 k8s 高可用集群部署就和小伙伴分享到这里，生活加油。
上面涉及到的 Ansible 剧本，静态pod yaml 文件，YUM配置文件，包括后来的一些常用工具安装的 yaml 文件都以整理上传 gitee， 小伙伴可自行下载，篇幅问题，没有展示。
获取git 仓库地址方式，关注 公总好 山河已无恙，回复 k8s-ha-deploy 即可获得地址。
博文部分内容参考
文中涉及参考链接内容版权归原作者所有，如有侵权请告知
https://kubernetes.io/zh-cn/docs/setup/production-environment/tools/
https://github.com/Mirantis/cri-dockerd
https://kubernetes.io/zh-cn/docs/setup/production-environment/container-runtimes/
https://github.com/kubernetes/kubeadm/blob/main/docs/ha-considerations.md#options-for-software-load-balancing
© 2018-2023 liruilonger@gmail.com, All rights reserved. 保持署名-非商用-相同方式共享(CC BY-NC-SA 4.0)