写在前面
- 实验需要一个
CNI为flannel的K8s集群 - 之前有一个
calico的版本有些旧了,所以国庆部署了一个 v1.31.1 版本 3 * master + 5 * work- 时间关系直接用的工具
kubespray - 博文内容为部署过程以及一些躺坑分享
- 需要科学上网
- 理解不足小伙伴帮忙指正 😃,生活加油
99%的焦虑都来自于虚度时间和没有好好做事，所以唯一的解决办法就是行动起来，认真做完事情，战胜焦虑，战胜那些心里空荡荡的时刻，而不是选择逃避。不要站在原地想象困难，行动永远是改变现状的最佳方式
机器准备
用到的机器
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$kubectl get nodes
NAME STATUS ROLES AGE VERSION
192.168.26.100 Ready control-plane 4h53m v1.31.1
192.168.26.101 Ready control-plane 4h49m v1.31.1
192.168.26.102 Ready control-plane 4h45m v1.31.1
192.168.26.103 Ready <none> 4h43m v1.31.1
192.168.26.105 Ready <none> 4h43m v1.31.1
192.168.26.106 Ready <none> 4h43m v1.31.1
192.168.26.107 Ready <none> 4h43m v1.31.1
192.168.26.108 Ready <none> 4h43m v1.31.1
┌──[root@liruilongs.github.io]-[~/kubespray]
└─
还需要一台客户机（liruilongs.github.io），也可以直接使用 集群 中的一台机器。建议单独出来，需要装一些 docker 之类的工具，放到一起可能会导致 runtime socket 污染。之前遇到一次，kubelet 运行要指定 runtime socket 的才行
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$hostnamectlStatic hostname: liruilongs.github.ioIcon name: computer-vmChassis: vm 🖴Machine ID: d8b7006e0b5246d3940d0da0b5e4d792Boot ID: 9f978ed1303842f6b8685a36624e5843Virtualization: vmware
Operating System: openEuler 24.03 (LTS)Kernel: Linux 6.6.0-44.0.0.50.oe2403.x86_64Architecture: x86-64Hardware Vendor: VMware, Inc.Hardware Model: VMware Virtual Platform
Firmware Version: 6.00Firmware Date: Wed 2020-07-22Firmware Age: 4y 2month 1w 4d
使用的系统： openEuler 24.03 (LTS)要选择长期支持版本，镜像下载地址：
https://www.openeuler.org/zh/download/?version=openEuler%2024.03%20LTS
通过 VMware 创建一台机器，配置一下 root 用户和装机磁盘，直接下一步就OK
其他的直接克隆一下,克隆完机器，通过IP 配置脚本统一配置一下IP，主机名称 DNS 相关
IP 配置脚本
┌──[root@192.168.26.100]-[~]
└─$cat set.sh
#!/bin/bash
if [ $# -eq 0 ]; thenecho "usage: `basename $0` num"exit 1
fi
[[ $1 =~ ^[0-9]+$ ]]
if [ $? -ne 0 ]; thenecho "usage: `basename $0` 10~240"exit 1
ficat > /etc/sysconfig/network-scripts/ifcfg-ens33 <<EOF
TYPE=Ethernet
PROXY_METHOD=none
BROWSER_ONLY=no
BOOTPROTO=none
IPADDR=192.168.26.${1}
PREFIX=24
DEFROUTE=yes
IPV4_FAILURE_FATAL=no
IPV6INIT=yes
IPV6_AUTOCONF=yes
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6_ADDR_GEN_MODE=default
NAME=ethernet-ens33
DEVICE=ens33
ONBOOT=yes
GATEWAY=192.168.26.2
DNS1=192.168.26.2
EOFsystemctl restart NetworkManager &> /dev/null
ip=$(ifconfig ens33 | awk '/inet /{print $2}')
sed -i '/192/d' /etc/issue
echo $ip
echo $ip >> /etc/issue
hostnamectl set-hostname vms${1}.liruilongs.github.io
echo "192.168.26.${1} vms${1}.liruilongs.github.io vms${1}" >> /etc/hostsecho '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
reboot
┌──[root@192.168.26.100]-[~]
└─$
运行方式,后面跟IP的 主机地址，每台机器运行一下
./set.sh 100
配置完成之后我们在客户机上面配置一下免密，当然这个不是必须
配置免密的IP列表
┌──[root@liruilongs.github.io]-[~]
└─$cat host_list
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
执行的免密脚本,需要装一个 expect， 脚本中 redhat 为 root 密码，需要修改为自己的
#!/bin/bash#@File : mianmi.sh
#@Time : 2022/08/20 17:45:53
#@Author : Li Ruilong
#@Version : 1.0
#@Desc : yum install -y expect
#@Contact : 1224965096@qq.com/usr/bin/expect <<-EOF
spawn ssh-keygen
expect "(/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 eof
EOFfor IP in $( cat host_list )
doif [ -n IP ];then/usr/bin/expect <<-EOF
spawn ssh-copy-id root@$IPexpect {"*yes/no*" { send "yes\r"}"*password*" { send "redhat\r" }}
expect {"*password" { send "redhat\r"}"#" { send "\r"}}
expect "#"
expect eof
EOF
fidone到这里就OK了，剩下的我们通过 ansible 来处理
kubespray 认识
使用的项目： https://github.com/kubernetes-sigs/kubespray
感兴趣小伙伴可以了解一下，这里我们 fork 到自己的仓库，方便后期更改, clone 一下
┌──[root@liruilongs.github.io]-[~]
└─$git clone https://github.com/LIRUILONGS/kubespray.git
然后根据文档描述来部署就OK
# Copy ``inventory/sample`` as ``inventory/mycluster``
cp -rfp inventory/sample inventory/mycluster# Update Ansible inventory file with inventory builder
declare -a IPS=(10.10.1.3 10.10.1.4 10.10.1.5)
CONFIG_FILE=inventory/mycluster/hosts.yaml python3 contrib/inventory_builder/inventory.py ${IPS[@]}# Review and change parameters under ``inventory/mycluster/group_vars``
cat inventory/mycluster/group_vars/all/all.yml
cat inventory/mycluster/group_vars/k8s_cluster/k8s-cluster.yml# Clean up old Kubernetes cluster with Ansible Playbook - run the playbook as root
# The option `--become` is required, as for example cleaning up SSL keys in /etc/,
# uninstalling old packages and interacting with various systemd daemons.
# Without --become the playbook will fail to run!
# And be mind it will remove the current kubernetes cluster (if it's running)!
ansible-playbook -i inventory/mycluster/hosts.yaml --become --become-user=root reset.yml# Deploy Kubespray with Ansible Playbook - run the playbook as root
# The option `--become` is required, as for example writing SSL keys in /etc/,
# installing packages and interacting with various systemd daemons.
# Without --become the playbook will fail to run!
ansible-playbook -i inventory/mycluster/hosts.yaml --become --become-user=root cluster.yml
简单看一下，项目比较重，大概 1173 个任务，实际执行的大概 500 左右，有些是系统兼容处理，辅助任务。
按照上面的要求 把给的 Demo 剧本相关的主机清单以及清单变量复制一份，修改 mycluster 为自己的名字 liruilong-cluster
需要的 ansible 版本比较新，当前机器的版本有些旧，所以我们使用 容器的方式,有个 ansible_version.yml 的剧本专门做 ansible 校验，可以通过排除 check 标签来跳过，但是试了下不太行，所以还是用容器了
┌──[root@liruilongs.github.io]-[~]
└─$docker run --rm -it -v /root/kubespray/:/kubespray/ -v "${HOME}"/.ssh/id_rsa:/root/.ssh/id_rsa quay.io/kubespray/kubespray:v2.26.0 bash
这里有个
坑，目录映射要直接映射整个项目，容器版本里面的项目有些旧了，和当前项目剧本不一致，执行了好多次发现，项目中给的方式是只映射了清单文件的目录。把整个项目映射过去，不要就映射清单文件目录,在说一遍
下面是项目中给的 容器方式运行的 Demo
git checkout v2.26.0
docker pull quay.io/kubespray/kubespray:v2.26.0
docker run --rm -it --mount type=bind,source="$(pwd)"/inventory/sample,dst=/inventory \--mount type=bind,source="${HOME}"/.ssh/id_rsa,dst=/root/.ssh/id_rsa \quay.io/kubespray/kubespray:v2.26.0 bash
# Inside the container you may now run the kubespray playbooks:
ansible-playbook -i /inventory/inventory.ini --private-key /root/.ssh/id_rsa cluster.yml
主机清单以及变量配置
在执行剧本之前我们需要处理下主机清单，以及清单变量，直接在客户机修改即可
主机清单配置
可以看到基本的分组：
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$cat inventory.ini
# ## Configure 'ip' variable to bind kubernetes services on a
# ## different ip than the default iface
# ## We should set etcd_member_name for etcd cluster. The node that is not a etcd member do not need to set the value, or can set the empty string value.
[all]
# node1 ansible_host=95.54.0.12 # ip=10.3.0.1 etcd_member_name=etcd1
# node2 ansible_host=95.54.0.13 # ip=10.3.0.2 etcd_member_name=etcd2
# node3 ansible_host=95.54.0.14 # ip=10.3.0.3 etcd_member_name=etcd3
# node4 ansible_host=95.54.0.15 # ip=10.3.0.4 etcd_member_name=etcd4
# node5 ansible_host=95.54.0.16 # ip=10.3.0.5 etcd_member_name=etcd5
# node6 ansible_host=95.54.0.17 # ip=10.3.0.6 etcd_member_name=etcd6# ## configure a bastion host if your nodes are not directly reachable
# [bastion]
# bastion ansible_host=x.x.x.x ansible_user=some_user[kube_control_plane]
# node1
# node2
# node3[etcd]
# node1
# node2
# node3[kube_node]
# node2
# node3
# node4
# node5
# node6
Demo 中的有些清单变量的配置，实际上这里的变量优先级要高于角色中的变量，配置重名会覆盖掉角色的变量，所以这里我们只配置主机，关于清单变量小伙伴可以看我之前的文章，有详细的优先级分析。
下面为我们配置的,跳过了 104 ，不吉利 ^_^
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat inventory/liruilong-cluster/inventory.ini
[all]
192.168.26.[101:103]
192.168.26.[105:108][kube_control_plane]
192.168.26.[100:102][etcd]
192.168.26.[100:102][kube_node]
192.168.26.103
192.168.26.[105:108]
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
通过 --graph 命令检查一下分组
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible-inventory -i inventory/liruilong-cluster/inventory.ini --graph
@all:|--@etcd:| |--192.168.26.100| |--192.168.26.101| |--192.168.26.102|--@kube_control_plane:| |--192.168.26.100| |--192.168.26.101| |--192.168.26.102|--@kube_node:| |--192.168.26.103| |--192.168.26.105| |--192.168.26.106| |--192.168.26.107| |--192.168.26.108|--@ungrouped:
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
我们配置了 root 免密，所以可以直接执行，没配置可能需要在 ansible 配置文件或者命令行做提权
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m ping -i inventory/liruilong-cluster/inventory.ini
192.168.26.103 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.105 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.101 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.102 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.106 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.107 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.108 | SUCCESS => {"changed": false,"ping": "pong"
}
192.168.26.100 | SUCCESS => {"changed": false,"ping": "pong"
}
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
下面我们看下清单变量的处理
CNI 插件修改
修改一下网络插件，前面有讲，我们CNI 是要用 flannel ，所以这里修改，如果就使用 calico，那么这里不需要动
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw ./ ../../roles/ -i -e "kube_network_plugin:"
./group_vars/k8s_cluster/k8s-cluster.yml:70:kube_network_plugin: calico
../../roles/container-engine/gvisor/molecule/default/prepare.yml:21: kube_network_plugin: cni
../../roles/container-engine/containerd/molecule/default/prepare.yml:26: kube_network_plugin: cni
../../roles/container-engine/kata-containers/molecule/default/prepare.yml:21: kube_network_plugin: cni
../../roles/container-engine/youki/molecule/default/prepare.yml:21: kube_network_plugin: cni
../../roles/container-engine/cri-dockerd/molecule/default/prepare.yml:21: kube_network_plugin: cni
../../roles/container-engine/cri-o/molecule/default/prepare.yml:26: kube_network_plugin: cni
../../roles/kubespray-defaults/defaults/main/main.yml:189:kube_network_plugin: calico
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$
需要关注下下面两个
./group_vars/k8s_cluster/k8s-cluster.yml:70:kube_network_plugin: calico
../../roles/kubespray-defaults/defaults/main/main.yml:189:kube_network_plugin: calico
这里我们简单看下 ansible 变量优先级，当使用多种方式定义相同变量时，Ansible将使用优先级规则为变量选取值。以下讨论优先级从低到高：
- 配置文件(ansible.cfg)
- 命令行选项
- 角色defaults变量
- host和group变量
- Play变量
- Extra变量(全局变量)
所以group_vars 的优先级要大于 角色中的defaults 目录下的变量，直接修改第一个就可以
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$vim +70 ./group_vars/k8s_cluster/k8s-cluster.yml
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw ./ ../../roles/ -i -e "kube_network_plugin:"
./group_vars/k8s_cluster/k8s-cluster.yml:70:kube_network_plugin: flannel
......
NTP 检查
新装的机器，时间同步检查，时区确认
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$timedatectl statusLocal time: 三 2024-10-02 21:02:09 CSTUniversal time: 三 2024-10-02 13:02:09 UTCRTC time: 三 2024-10-02 19:04:08Time zone: Asia/Shanghai (CST, +0800)
System clock synchronized: noNTP service: activeRTC in local TZ: no
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$
时区是OK，但是时间不同步
剧本中有相关配置，直接两个都改成 true
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw . -i -C 2 -e "ntp"
./group_vars/all/all.yml-123-# kube_webhook_token_auth_ca_data: "LS0t..."
./group_vars/all/all.yml-124-
./group_vars/all/all.yml:125:## NTP Settings
./group_vars/all/all.yml-126-# Start the ntpd or chrony service and enable it at system boot.
./group_vars/all/all.yml-127-ntp_enabled: false
./group_vars/all/all.yml-128-ntp_manage_config: false
./group_vars/all/all.yml-129-ntp_servers:
./group_vars/all/all.yml:130: - "0.pool.ntp.org iburst"
./group_vars/all/all.yml:131: - "1.pool.ntp.org iburst"
./group_vars/all/all.yml:132: - "2.pool.ntp.org iburst"
./group_vars/all/all.yml:133: - "3.pool.ntp.org iburst"
./group_vars/all/all.yml-134-
./group_vars/all/all.yml-135-## Used to control no_log attribute
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$
./group_vars/all/all.yml-127-ntp_enabled: true
./group_vars/all/all.yml-128-ntp_manage_config: true
确认 NTP 服务器确实可达
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$ ntpdate -q 0.pool.ntp.org
server 202.118.1.81, stratum 1, offset +45688.212514, delay 0.08925
server 193.182.111.143, stratum 2, offset +45688.206599, delay 0.19637
server 193.182.111.14, stratum 2, offset +45688.191100, delay 0.19890
server 78.46.102.180, stratum 2, offset +45688.213370, delay 0.304822 Oct 21:08:31 ntpdate[5225]: step time server 202.118.1.81 offset +45688.212514 sec
这里又遇到一个
坑，配置之后看上去是生效了，但是时间差的太远了，在执行剧本的时候，有个生成 tocken 的步骤，死活生成不了,怀疑是 NTP 的问题，后来通过 ansible 手动同步了一遍可以了
检查 chronyd 服务状态
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a "systemctl is-active chronyd" -i inventory/liruilong-cluster/inventory.ini
192.168.26.102 | CHANGED | rc=0 >>
active
192.168.26.103 | CHANGED | rc=0 >>
active
192.168.26.106 | CHANGED | rc=0 >>
active
192.168.26.105 | CHANGED | rc=0 >>
active
192.168.26.101 | CHANGED | rc=0 >>
active
192.168.26.108 | CHANGED | rc=0 >>
active
192.168.26.107 | CHANGED | rc=0 >>
active
192.168.26.100 | CHANGED | rc=0 >>
active
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
设置当前时间
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a " date -s '2024-10-05 15:41'" -i inventory/liruilong-cluster/inventory.ini
这里实际上还需要系统时间设置BIOS时间：hwclock -w
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a "hwclock -w" -i inventory/liruilong-cluster/inventory.ini
同步检查
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a "chronyc tracking" -i inventory/liruilong-cluster/inventory.ini
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a "chronyc sources -v" -i inventory/liruilong-cluster/inventory.ini
防火墙处理
项目中说防火墙不受管理，您需要像以前一样实施自己的规则。 为了避免在部署过程中出现任何问题，您应该禁用防火墙
所以这里我们直接禁用，stop 之后 disable 一下，下面为检查的命令
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$ansible all -m shell -a "systemctl is-active firewalld" -i inventory/liruilong-cluster/inventory.ini
当然不管也可以，可以通过修改 zone 实现，下面为 剧本中使用 firewalld 模块Demo
# 关闭防火墙,这里设置为 trusted ,以后可能处理漏洞使用
- name: firewalld setting trustedfirewalld:zone: trustedpermanent: yesstate: enabled
SELinux
确认是否有相关配置
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw ./ ../../roles/ -i -e "selinux"
......................
../../roles/kubespray-defaults/defaults/main/main.yml:6:# selinux state
....................
可以看到在 kubespray-defaults 角色中有对 SELinux 做配置 preinstall_selinux_state: permissive
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$vim +6 ../../roles/kubespray-defaults/defaults/main/main.yml
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw ./ ../../roles/ -i -e "preinstall_selinux_state:"
../../roles/kubespray-defaults/defaults/main/main.yml:7:preinstall_selinux_state: permissive
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$
这里不需要修改，kubespray-defaults 角色在任务中多次被调用，这里小伙伴可以通过 核心剧本看到。
当然，你也可以通过剧本中 SELinux 模块来关闭它
# 关闭 SELinux
- name: Disable SELinuxselinux:state: disabled
DNS
DNS 这里也有一个坑， 配置了没有生效，，后面通过ansible 同步了一遍
/etc/resolv.conf.下面是配置项
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat inventory/liruilong-cluster/group_vars/all/all.yml | grep -i -C 2 dns## By default, Kubespray collects nameservers on the host. It then adds the previously collected nameservers in nameserverentries.
## If true, Kubespray does not include host nameservers in nameserverentries in dns_late stage. However, It uses the nameserver to make sure cluster installed safely in dns_early stage.
## Use this option with caution, you may need to define your dns servers. Otherwise, the outbound queries such as www.google.com may fail.
# disable_host_nameservers: false## Upstream dns servers
# upstream_dns_servers:
# - 8.8.8.8
# - 8.8.4.4
这是修改完的配置
┌──[root@liruilongs.github.io]-[~/kubespray/inventory]
└─$grep -rnw ./ -i -A 4 -e "upstream_dns_servers"./liruilong-cluster/group_vars/all/all.yml:39:upstream_dns_servers:
./liruilong-cluster/group_vars/all/all.yml-40- - 8.8.8.8
./liruilong-cluster/group_vars/all/all.yml-41- - 8.8.4.4
./liruilong-cluster/group_vars/all/all.yml-42- - 114.114.114.114
./liruilong-cluster/group_vars/all/all.yml-43-
┌──[root@liruilongs.github.io]-[~/kubespray/inventory]
└─$
但是在后来 ctr 拉集群相关的镜像的时候,发现 DNS 解析不了 registry.k8s.io，看了 /etc/resolv.conf 才发现什么都没配置，所以通过 ansible 同步了一遍 /etc/resolv.conf.
下面为同步的文件,直接通过 ansible 的 copy 模块同步即可
# Generated by NetworkManager
nameserver 192.168.26.2
nameserver 8.8.8.8
nameserver 8.8.4.4
nameserver 114.114.114.114
options ndots:2 timeout:2 attempts:2
镜像仓库修改
国内 docker 镜像仓库拉不了，但是不知道为什么，我用了 VPN 还是不行，各种加速器试了好久，阿里的加速器不能免费用了。最后找个一个开源，dockerpull.com ，很感谢维护者，我折腾了好久，有条件的小伙伴可以捐赠下作者。
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$grep -rnw ./ ../../roles/ -i -e "docker_image_repo"
..............................................
../../roles/kubernetes-apps/external_cloud_controller/hcloud/templates/external-hcloud-cloud-controller-manager-ds.yml.j2:35: - image: {{ docker_image_repo }}/hetznercloud/hcloud-cloud-controller-manager:{{ external_hcloud_cloud.controller_image_tag }}
../../roles/kubernetes-apps/external_cloud_controller/hcloud/templates/external-hcloud-cloud-controller-manager-ds-with-networks.yml.j2:35: - image: {{ docker_image_repo }}/hetznercloud/hcloud-cloud-controller-manager:{{ external_hcloud_cloud.controller_image_tag }}
../../roles/kubespray-defaults/defaults/main/download.yml:94:docker_image_repo: "docker.io"
../../roles/kubespray-defaults/defaults/main/download.yml:240:flannel_image_repo: "{{ docker_image_repo }}/flannel/flannel"
../../roles/kubespray-defaults/defaults/main/download.yml:242:flannel_init_image_repo: "{{ docker_image_repo }}/flannel/flannel-cni-plugin"
../../roles/kubespray-defaults/defaults/main/download.yml:259:netcheck_agent_image_repo: "{{ docker_image_repo }}/mirantis/k8s-netchecker-agent"
........................................
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$
修改位置：
../../roles/kubespray-defaults/defaults/main/download.yml:94:docker_image_repo: "docker.io"
# docker image repo define
#docker_image_repo: "docker.io"
docker_image_repo: "dockerpull.com"# quay image repo define
quay_image_repo: "quay.io"# github image repo define (ex multus only use that)
github_image_repo: "ghcr.io"
运行剧本
┌──[root@liruilongs.github.io]-[~]
└─$docker run --rm -it -v /root/kubespray/:/kubespray/ -v "${HOME}"/.ssh/id_rsa:/root/.ssh/id_rsa quay.io/kubespray/kubespray:v2.26.0 bash
root@962b70466dd9:/kubespray# ansible-playbook -i inventory/liruilong-cluster/inventory.ini cluster.yml -f 8
这里需要注意，因为科学上网的原因，可能需要多次运行，剧本中途会挂掉，下面为执行成功的输出
...........................................
PLAY RECAP ******************************************************************************************************************************************************************************************************
192.168.26.100 : ok=515 changed=54 unreachable=0 failed=0 skipped=934 rescued=0 ignored=2
192.168.26.101 : ok=487 changed=49 unreachable=0 failed=0 skipped=900 rescued=0 ignored=2
192.168.26.102 : ok=486 changed=50 unreachable=0 failed=0 skipped=898 rescued=0 ignored=2
192.168.26.103 : ok=410 changed=14 unreachable=0 failed=0 skipped=667 rescued=0 ignored=1
192.168.26.105 : ok=388 changed=10 unreachable=0 failed=0 skipped=597 rescued=0 ignored=1
192.168.26.106 : ok=388 changed=10 unreachable=0 failed=0 skipped=597 rescued=0 ignored=1
192.168.26.107 : ok=388 changed=10 unreachable=0 failed=0 skipped=597 rescued=0 ignored=1
192.168.26.108 : ok=388 changed=10 unreachable=0 failed=0 skipped=597 rescued=0 ignored=1Friday 04 October 2024 22:15:07 +0000 (0:00:00.843) 0:33:16.408 ********
===============================================================================
.............................................
kubernetes/control-plane : Joining control plane node to the cluster. ---------------------------------------------------------------------------------------------------------------------------------- 494.78s
kubernetes/control-plane : Kubeadm | Initialize first control plane node ------------------------------------------------------------------------------------------------------------------------------- 244.24s
kubernetes/control-plane : Install script to renew K8S control plane certificates ----------------------------------------------------------------------------------------------------------------------- 50.58s
etcd : Gen_certs | Write etcd member/admin and kube_control_plane client certs to other etcd nodes ------------------------------------------------------------------------------------------------------ 36.03s
etcd : Gen_certs | Write node certs to other etcd nodes ------------------------------------------------------------------------------------------------------------------------------------------------- 26.20s
kubernetes/kubeadm : Join to cluster -------------------------------------------------------------------------------------------------------------------------------------------------------------------- 21.51s
network_plugin/cni : CNI | Copy cni plugins ------------------------------------------------------------------------------------------------------------------------------------------------------------- 20.59s
kubernetes-apps/ansible : Kubernetes Apps | Lay Down CoreDNS templates ---------------------------------------------------------------------------------------------------------------------------------- 19.91s
container-engine/containerd : Containerd | Unpack containerd archive ------------------------------------------------------------------------------------------------------------------------------------ 17.56s
kubernetes/control-plane : Control plane | Remove scheduler container containerd/crio ------------------------------------------------------------------------------------------------------------------- 14.68s
container-engine/crictl : Extract_file | Unpacking archive ---------------------------------------------------------------------------------------------------------------------------------------------- 14.62s
container-engine/containerd : Download_file | Download item --------------------------------------------------------------------------------------------------------------------------------------------- 13.04s
container-engine/runc : Download_file | Download item --------------------------------------------------------------------------------------------------------------------------------------------------- 13.02s
container-engine/nerdctl : Extract_file | Unpacking archive --------------------------------------------------------------------------------------------------------------------------------------------- 12.99s
kubernetes-apps/ansible : Kubernetes Apps | Start Resources --------------------------------------------------------------------------------------------------------------------------------------------- 12.54s
container-engine/crictl : Download_file | Download item ------------------------------------------------------------------------------------------------------------------------------------------------- 12.43s
container-engine/nerdctl : Download_file | Download item ------------------------------------------------------------------------------------------------------------------------------------------------ 11.80s
kubernetes/control-plane : Fixup kubelet client cert rotation 2/2 --------------------------------------------------------------------------------------------------------------------------------------- 10.29s
kubernetes-apps/network_plugin/flannel : Flannel | Wait for flannel subnet.env file presence ------------------------------------------------------------------------------------------------------------- 9.50s
download : Extract_file | Unpacking archive -------------------------------------------------------------------------------------------------------------------------------------------------------------- 8.52s
root@962b70466dd9:/kubespray#
然后就可以在 master 上面看到集群信息了
┌──[root@192.168.26.100]-[/]
└─$kubectl get nodes
NAME STATUS ROLES AGE VERSION
192.168.26.100 Ready control-plane 31m v1.31.1
192.168.26.101 Ready control-plane 28m v1.31.1
192.168.26.102 Ready control-plane 23m v1.31.1
192.168.26.103 Ready <none> 21m v1.31.1
192.168.26.105 Ready <none> 21m v1.31.1
192.168.26.106 Ready <none> 21m v1.31.1
192.168.26.107 Ready <none> 21m v1.31.1
192.168.26.108 Ready <none> 21m v1.31.1
┌──[root@192.168.26.100]-[/]
└─$
到这里我们就完成了安装，在客户机上面拷贝一下 kubeconfig 文件和kubectl 工具，我们可以直接在 在客户机操作
┌──[root@192.168.26.100]-[~]
└─$kubectl version
Client Version: v1.31.1
Kustomize Version: v5.4.2
Server Version: v1.31.1
┌──[root@192.168.26.100]-[~]
└─$
┌──[root@192.168.26.100]-[/usr/local/bin]
└─$scp ./kubectl root@192.168.26.131:/usr/local/binAuthorized users only. All activities may be monitored and reported.
root@192.168.26.131's password:
kubectl 100% 54MB 249.9MB/s 00:00
这里在安装完之前，发现家里的机器 CPU 核心，能效核饱和，性能核空闲状态,集群卡的命令都运行不了
解决办法： 修改电源计划为性能模式+管理员方式运行
这是因为在Intel酷睿12/13代处理器中，引入了一种新的大小核架构，通常称为“混合核心”架构。这种架构在同一芯片上集成了性能核（大核）和能效核（小核）。
- 性能核用于处理高负载、高性能要求的任务
- 能效核则用于处理低负载、低功耗要求的任务
我的机器 Intel Core i7-12700 处理器实际上具有以下线程配置：
性能核（Performance Cores）：
- 数量：8个
- 每个核心支持的线程数：2
- 总线程数（仅性能核）：8 × 2 = 16线程
能效核（Efficient Cores）：
- 数量：4个
- 每个核心支持的线程数：1
- 总线程数（仅能效核）：4 × 1 = 4线程
相关工具的安装
前面我们安装没有安装任何工具，默认情况下这些都是不安装的，比如 HELM，krew 之类，这些工具也可用通过角色的方式安装
┌──[root@liruilongs.github.io]-[~/kubespray/roles/kubernetes-apps]
└─$ls
ansible container_engine_accelerator external_provisioner krew metrics_server policy_controller
argocd container_runtimes gateway_api kubelet-csr-approver network_plugin registry
cloud_controller csi_driver helm meta node_feature_discovery scheduler_plugins
cluster_roles external_cloud_controller ingress_controller metallb persistent_volumes snapshots
┌──[root@liruilongs.github.io]-[~/kubespray/roles/kubernetes-apps]
└─$
helm、argocd 等工具安装
修改下面的文件，重新执行剧本即可
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat ./inventory/liruilong-cluster/group_vars/k8s_cluster/addons.yml
下面为修改之后的配置项
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat ./inventory/liruilong-cluster/group_vars/k8s_cluster/addons.yml | grep -v ^# | grep -v ^$
---
helm_enabled: true
registry_enabled: true
metrics_server_enabled: true
local_path_provisioner_enabled: true
local_volume_provisioner_enabled: true
cephfs_provisioner_enabled: true
rbd_provisioner_enabled: true
gateway_api_enabled: true
ingress_nginx_enabled: true
ingress_publish_status_address: ""
ingress_alb_enabled: true
cert_manager_enabled: true
metallb_enabled: true
metallb_speaker_enabled: "{{ metallb_enabled }}"
metallb_namespace: "metallb-system"
metallb_version: v0.13.9
metallb_protocol: "layer2"
metallb_port: "7472"
metallb_memberlist_port: "7946"
argocd_enabled: true
krew_enabled: true
krew_root_dir: "/usr/local/krew"
kube_vip_enabled: true
kube_vip_arp_enabled: true
kube_vip_controlplane_enabled: true
kube_vip_address: 192.168.26.120
loadbalancer_apiserver:address: "{{ kube_vip_address }}"port: 6443
node_feature_discovery_enabled: false
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
剧本执行的时候只执行 apps 的 标签即可
root@9d2aab6d0ea8:/kubespray# ansible-playbook -i inventory/liruilong-cluster/inventory.ini cluster.yml --tags apps
部分应用截图
metallb
软 LB metallb 安装,用 自带的角色安装 metallb 一直有报错，解决不了，这里手工安装
┌──[root@liruilongs.github.io]-[~/kubespray/inventory/liruilong-cluster]
└─$kubectl apply -f https://raw.githubusercontent.com/metallb/metallb/v0.13.9/config/manifests/metallb-native.yaml
分配 IP 池
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$vim pool.yaml
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat pool.yaml
apiVersion: metallb.io/v1beta1
kind: IPAddressPool
metadata:name: first-poolnamespace: metallb-system
spec:addresses:- 192.168.26.220-192.168.26.249
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$kubectl apply -f pool.yaml
ipaddresspool.metallb.io/first-pool created
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$vim l2a.yaml
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$cat l2a.yaml
apiVersion: metallb.io/v1beta1
kind: L2Advertisement
metadata:name: examplenamespace: metallb-system
┌──[root@liruilongs.github.io]-[~/kubespray]
└─$kubectl apply -f l2a.yaml
l2advertisement.metallb.io/example created
kube-prometheus-stack
prometheus 套件安装，需要先安装上面的 metrics_server
┌──[root@192.168.26.100]-[~]
└─$helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
"prometheus-community" has been added to your repositories
┌──[root@192.168.26.100]-[~]
└─$helm repo update
Hang tight while we grab the latest from your chart repositories...
...Successfully got an update from the "prometheus-community" chart repository
Update Complete. ⎈Happy Helming!⎈
┌──[root@192.168.26.100]-[~]
└─$helm install prometheus prometheus-community/kube-prometheus-stack
NAME: prometheus
LAST DEPLOYED: Sat Oct 5 20:25:46 2024
NAMESPACE: default
STATUS: deployed
REVISION: 1
NOTES:
kube-prometheus-stack has been installed. Check its status by running:kubectl --namespace default get pods -l "release=prometheus"Visit https://github.com/prometheus-operator/kube-prometheus for instructions on how to create & configure Alertmanager and Prometheus instances using the Operator.
关于 openEuler 24.03 (LTS) 部署 K8s(v1.31.1) 高可用集群(Kubespray Ansible 方式) 就和小伙伴分享到这里，之后有机会会分析一下Kubespray 集群剧本以及各个角色。
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://github.com/kubernetes-sigs/kubespray
© 2018-2024 liruilonger@gmail.com, 保持署名-非商用-相同方式共享(CC BY-NC-SA 4.0)
