从零开始基于Archlinux 安装 containerd + k8s

2023-04-27 11:59 由 Berkaroad 发表于 #其他

下载ISO文件：https://mirrors.tuna.tsinghua.edu.cn/archlinux/iso/latest/

k8s: v1.26.4；calico: 3.25.1

1. 准备工作

以虚拟机VMWare为例。

使用EFI 非默认BIOS启动。如果不使用EFI，那么后续安装引导时也使用非EFI。

Controller-Panel节点（master）

节点列表：

hostname ip

k8s-master1 10.0.2.101/24

k8s-master2 10.0.2.102/24

k8s-master3 10.0.2.103/24

CPU设置：2Core

内存设置：2GB

磁盘：20GB

网卡设置：网卡1（ens33）为自定义NAT
Worker节点

节点列表：

hostname ip

k8s-worker1 10.0.2.111/24

k8s-worker2 10.0.2.112/24

k8s-worker3 10.0.2.113/24

CPU设置：2Core

内存设置：4GB

磁盘：20GB

网卡设置：网卡1（ens33）为自定义NAT

hostname	ip
k8s-master1	10.0.2.101/24
k8s-master2	10.0.2.102/24
k8s-master3	10.0.2.103/24

hostname	ip
k8s-worker1	10.0.2.111/24
k8s-worker2	10.0.2.112/24
k8s-worker3	10.0.2.113/24

2. 磁盘管理

2.1 磁盘分区

使用GUID分区表，分2个区：

1）EFI System（EF00），Last sector： +500M （500MB）
2）Linux filesystem（8300），Last sector：<回车>（为剩余容量）

gdisk /dev/sda

2.2 磁盘格式化

mkfs.vfat -F32 /dev/sda1 # ESP分区 挂载 /boot
mkfs.ext4 /dev/sda2 # LFS分区 挂载 /

2.3 磁盘挂载

mount /dev/sda2 /mnt # 挂载root分区
mkdir /mnt/boot # 创建 /boot 目录
mount /dev/sda2 /mnt/boot # 挂载boot分区

lsblk # 查看分区挂载情况

3. 安装系统

3.1 安装系统文件

vim /etc/pacman.d/mirrorlist # 在顶部添加如下镜像服务器

Server = https://mirrors.tuna.tsinghua.edu.cn/archlinux/$repo/os/$arch
#Server = https://mirrors.aliyun.com/archlinux/$repo/os/$arch

# 安装系统
pacstrap /mnt base base-devel

3.2 配置fstab

genfstab -U /mnt > /mnt/etc/fstab # 生成分区挂载表

编辑 fstab

vim /mnt/etc/fstab
# SSD的追加options “discard,noatime”

3.3 配置系统

编辑 /mnt/etc/pacman.conf文件，加入下面的内容：

[archlinuxcn]
Server = https://mirrors.tuna.tsinghua.edu.cn/archlinuxcn/$arch
#Server = https://mirrors.aliyun.com/archlinuxcn/$arch

切换root目录到新系统

arch-chroot /mnt /bin/bash

现在可以全面升级系统：

pacman -Syy # 切换了root目录，因此需要重新更新软件包缓存
pacman -S archlinuxcn-keyring
pacman -S vim bash-completion yay fakeroot
ln -s /usr/bin/vim /usr/bin/vi

3.4 安装引导程序

# 安装linux内核
pacman -S linux-lts linux-firmware
# 安装 Micro Code
pacman -S amd-ucode # intel安装 intel-ucode

bootctl install # boot-loader

vim /boot/loader/entries/arch.conf
title Arch Linux
linux /vmlinuz-linux-lts
initrd /amd-ucode.img # intel的为 /intel-ucode.img
initrd /initramfs-linux-lts.img
options root=/dev/sda2 rw

vim /boot/loader/entries/arch-fallback.conf
title Arch Linux (fallback initramfs)
linux /vmlinuz-linux-lts
initrd /amd-ucode.img # intel的为 /intel-ucode.img
initrd /initramfs-linux-lts-fallback.img
options root=/dev/sda2 rw

vim /boot/efi/loader/loader.conf
default arch.conf
timeout 2
console-mode max
editor no

# 验证文件路径是否正确
bootctl list
bootctl status

3.5 安装OpenSSH

pacman -S openssh

sed -i 's/#PermitRootLogin\ prohibit-passwd/PermitRootLogin yes/g' /etc/ssh/sshd_config

systemctl enable sshd

3.6 主机名

echo <host-name> > /etc/hostname

3.7 设置root密码

passwd

3.8 网络配置

使用 systemd-networkd

VMWare 网络配置：
NAT模式
网段：10.0.2.0/24
DHCP：10.0.2.200 - 10.0.2.254
网关：10.0.2.2 （不要设置为10.0.2.1，否则会导致无法访问外网）

vim /etc/systemd/network/20-wired.network
[Match]
Name=ens33

[Network]
#DHCP=ipv4 # 使用dhcp时启用
Address=10.0.2.101/24
Gateway=10.0.2.2
DNS=223.5.5.5
DNS=223.6.6.6

systemctl enable systemd-networkd
systemctl enable systemd-resolved

3.9 重启系统，并从硬盘引导

exit # 退出chroot
reboot # 重启后重新引导进入已安装的系统

3.10 本地化配置

vim /etc/locale.gen

en_US.UTF-8 UTF-8
zh_CN.GBK GBK
zh_CN.UTF-8 UTF-8
zh_CN GB2312

locale-gen # 生成locale
echo 'LANG=en_US.UTF-8' > /etc/locale.conf # 设置默认的 locale

3.11 时区配置

ln -s /usr/share/zoneinfo/Asia/Shanghai /etc/localtime

3.12 硬件时间设置

# date -s '2022-7-5 16:49:45'
hwclock --systohc --utc #采用UTC，将系统时间写入硬件时钟
# hwclock --hctosys --utc #采用UTC，将硬件时钟写入系统时间

3.13 安装DNS服务

pacman -S bind
# 参见： https://wiki.archlinux.org/title/BIND

4. 安装k8s

使用kubeadm安装： https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/install-kubeadm/

pacman -S kubeadm kubelet kubectl containerd
systemctl enable containerd
systemctl start containerd
systemctl enable kubelet
systemctl start kubelet

4.1 配置containerd

创建 /etc/modules-load.d/containerd.conf 配置文件:

cat << EOF > /etc/modules-load.d/containerd.conf
overlay
br_netfilter
EOF

修改 containerd 配置

# 修改配置
mkdir -p /etc/containerd
if [ ! -f /etc/containerd/config.toml ]; then
    containerd config default > /etc/containerd/config.toml
fi

# 设置 systemd_cgroup 为 true
sed -i 's/SystemdCgroup = false/SystemdCgroup = true/g' /etc/containerd/config.toml
sed -i 's/k8s.gcr.io/registry.aliyuncs.com\/google_containers/g' /etc/containerd/config.toml

systemctl restart containerd

echo 'alias docker="crictl --runtime-endpoint unix:///var/run/containerd/containerd.sock"' > /etc/profile.d/containerd.sh
source /etc/profile.d/containerd.sh

# 确保containerd 的cgroup 为 SystemdCgroup
crictl --runtime-endpoint unix:///var/run/containerd/containerd.sock info | grep SystemdCgroup | awk -F ': ' '{ print $2 }'
true

4.2 拉取k8s镜像

通过参数 --image-repository 指定k8s镜像的仓库地址

kubeadm config images pull --image-repository=registry.aliyuncs.com/google_containers --kubernetes-version=v1.26.4

4.3 创建k8s集群

# 应搭建负载均衡后，使用负载均衡IP，此处用自建DNS服务来实现： 10.0.2.101 cluster.berkaroad.com

# 这个版本的kubelet，命令行参数 `--cni-bin-dir` 已经取消，因此需要拿掉此参数
sed -i 's/--cni-bin-dir=\/usr\/lib\/cni//g' /etc/kubernetes/kubelet.env

# 初始化k8s集群
kubeadm init  --image-repository=registry.aliyuncs.com/google_containers --kubernetes-version=v1.26.4 --control-plane-endpoint=cluster.berkaroad.com --apiserver-advertise-address=10.0.2.101 --pod-network-cidr=10.100.0.0/16 --service-cidr=10.101.0.0/16 --service-dns-domain=cluster.berkaroad.com --upload-certs --v=5

# 执行成功后，根据提示，配置
mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

# 注意：集群中时间必须保持一致，否则会加入集群失败，错误信息： x509: certificate has expired or is not yet valid: current time 2022-07-05T03:57:41+08:00 is before 2022-07-04T23:42:18Z

# You can now join any number of the control-plane node running the following command on each as root:
kubeadm join cluster.berkaroad.com:6443 --token v3e3b4.a52hqkbd1rlxgkun \
    --discovery-token-ca-cert-hash sha256:877bc4de6051c6aee8401bb99e6a3114f6d5a5fa7d87131c0b6377ce2419e5a3 \
    --control-plane --certificate-key 6b6050b43696814460032c521569377829e6bda6d39ac69e1d650d5bfdad1a44

# 如果 --certificate-key 过期了，执行如下：
kubeadm init phase upload-certs --upload-certs


# Then you can join any number of worker nodes by running the following on each as root:
kubeadm join cluster.berkaroad.com:6443 --token v3e3b4.a52hqkbd1rlxgkun \
    --discovery-token-ca-cert-hash sha256:877bc4de6051c6aee8401bb99e6a3114f6d5a5fa7d87131c0b6377ce2419e5a3

# 如果token过期了，可以执行如下：
kubeadm token create --print-join-command


# 如果失败，检查 cgroup 是否一致（docker或者containerd 和 kubelet）
# 查看 kubeadm 使用的 CRI 为 containerd 还是 docker
cat /var/lib/kubelet/kubeadm-flags.env
KUBELET_KUBEADM_ARGS="--container-runtime=remote --container-runtime-endpoint=unix:///var/run/containerd/containerd.sock --pod-infra-container-image=registry.aliyuncs.com/google_containers/pause:3.9"

# 查看 kubelet 的 cgroup driver
cat /var/lib/kubelet/config.yaml | grep cgroupDriver | awk -F ': ' '{ print $2 }'
systemd

4.4 加入control-plane节点

# 应搭建负载均衡后，使用负载均衡IP
echo '10.0.2.101 cluster.berkaroad.com' >> /etc/hosts

# 这个版本的kubelet，命令行参数 `--cni-bin-dir` 已经取消，因此需要拿掉此参数
sed -i 's/--cni-bin-dir=\/usr\/lib\/cni//g' /etc/kubernetes/kubelet.env


# 注意：集群中时间必须保持一致，否则会加入集群失败，错误信息： x509: certificate has expired or is not yet valid: current time 2022-07-05T03:57:41+08:00 is before 2022-07-04T23:42:18Z

# You can now join any number of the control-plane node running the following command on each as root:
kubeadm join cluster.berkaroad.com:6443 --token v3e3b4.a52hqkbd1rlxgkun \
    --discovery-token-ca-cert-hash sha256:877bc4de6051c6aee8401bb99e6a3114f6d5a5fa7d87131c0b6377ce2419e5a3 \
    --control-plane --certificate-key 6b6050b43696814460032c521569377829e6bda6d39ac69e1d650d5bfdad1a44

# 如果 --certificate-key 过期了，执行如下：
kubeadm init phase upload-certs --upload-certs

# 如果token过期了，可以执行如下：
kubeadm token create --print-join-command

# 执行成功后，根据提示，配置
mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

4.5 加入worker节点

# 应搭建负载均衡后，使用负载均衡IP
echo '10.0.2.101 cluster.berkaroad.com' >> /etc/hosts

# 这个版本的kubelet，命令行参数 `--cni-bin-dir` 已经取消，因此需要拿掉此参数
sed -i 's/--cni-bin-dir=\/usr\/lib\/cni//g' /etc/kubernetes/kubelet.env


# 执行成功后，根据提示，配置
mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

# 注意：集群中时间必须保持一致，否则会加入集群失败，错误信息： x509: certificate has expired or is not yet valid: current time 2022-07-05T03:57:41+08:00 is before 2022-07-04T23:42:18Z

# Then you can join any number of worker nodes by running the following on each as root:
kubeadm join cluster.berkaroad.com:6443 --token v3e3b4.a52hqkbd1rlxgkun \
    --discovery-token-ca-cert-hash sha256:877bc4de6051c6aee8401bb99e6a3114f6d5a5fa7d87131c0b6377ce2419e5a3

# 如果token过期了，可以执行如下：
kubeadm token create --print-join-command

4.6 安装CNI：Calico

kubectl apply -f https://projectcalico.docs.tigera.io/archive/v3.25/manifests/calico.yaml

4.7 查看k8s集群

节点信息：

kubectl get no -o wide
NAME          STATUS   ROLES           AGE   VERSION   INTERNAL-IP   EXTERNAL-IP   OS-IMAGE     KERNEL-VERSION   CONTAINER-RUNTIME
k8s-master1   Ready    control-plane   23m   v1.26.3   10.0.2.101    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0
k8s-master2   Ready    control-plane   22m   v1.26.3   10.0.2.102    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0
k8s-master3   Ready    control-plane   22m   v1.26.3   10.0.2.103    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0
k8s-worker1   Ready    <none>          20m   v1.26.3   10.0.2.111    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0
k8s-worker2   Ready    <none>          18m   v1.26.3   10.0.2.112    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0
k8s-worker3   Ready    <none>          17m   v1.26.3   10.0.2.113    <none>        Arch Linux   6.1.25-1-lts     containerd://1.7.0

pod信息：

kubectl get po -n kube-system
NAME                                      READY   STATUS    RESTARTS      AGE
calico-kube-controllers-57b57c56f-g62jv   1/1     Running   0             6m5s
calico-node-2b5f9                         1/1     Running   0             6m5s
calico-node-flbmt                         1/1     Running   0             6m5s
calico-node-hwtvh                         1/1     Running   0             6m5s
calico-node-j6dkp                         1/1     Running   0             6m5s
calico-node-jqcfg                         1/1     Running   0             6m5s
calico-node-lrq7q                         1/1     Running   0             6m5s
coredns-5bbd96d687-fd9j7                  1/1     Running   0             24m
coredns-5bbd96d687-kd48v                  1/1     Running   0             24m
etcd-k8s-master1                          1/1     Running   0             25m
etcd-k8s-master2                          1/1     Running   0             23m
etcd-k8s-master3                          1/1     Running   0             23m
kube-apiserver-k8s-master1                1/1     Running   0             25m
kube-apiserver-k8s-master2                1/1     Running   0             23m
kube-apiserver-k8s-master3                1/1     Running   0             23m
kube-controller-manager-k8s-master1       1/1     Running   0             25m
kube-controller-manager-k8s-master2       1/1     Running   0             23m
kube-controller-manager-k8s-master3       1/1     Running   0             22m
kube-proxy-6v7b9                          1/1     Running   0             18m
kube-proxy-7dnmx                          1/1     Running   0             22m
kube-proxy-c2cdd                          1/1     Running   0             23m
kube-proxy-k4l4c                          1/1     Running   0             19m
kube-proxy-rjw8j                          1/1     Running   0             24m
kube-proxy-zrcvw                          1/1     Running   0             23m
kube-scheduler-k8s-master1                1/1     Running   0             25m
kube-scheduler-k8s-master2                1/1     Running   0             23m
kube-scheduler-k8s-master3                1/1     Running   0             23m

附录

包签名错误

error: libcap: signature from "David Runge <[email protected]>" is marginal trust
:: File /var/cache/pacman/pkg/libcap-2.65-1-x86_64.pkg.tar.zst is corrupted (invalid or corrupted package (PGP signature)).
Do you want to delete it? [Y/n] Y
error: failed to commit transaction (invalid or corrupted package)
Errors occurred, no packages were upgraded.

更新pacman key证书

pacman -S gnupg
pacman -Sy archlinux-keyring
pacman-key --populate archlinux
pacman-key --refresh-keys
pacman -Syux