【K3s 系列 01】从零搭建 K3s

Mon, 29 Jun 2026 09:19:11 +0000

一、获取系统基础硬件信息

#基础架构（输出参考：如果是 x86_64 或 aarch64，即可完美支持。） pig@vm-debian-13:~$ uname -m x86_64


#处理器 （K3s 单节点极其轻量，最低仅需 1核 CPU 和 512MB 内存（推荐 1核1G 以上）。）
pig@vm-debian-13:~$ lscpu | grep -E “Model name|CPU(s):”
CPU(s):                                  2
Model name:                              13th Gen Intel(R) Core(TM) i5-13400
NUMA node0 CPU(s):                       0,1

#内存大小 pig@vm-debian-13:~$ free -m total used free shared buff/cache available Mem: 3883 886 1917 0 1334 2997 Swap: 0 0

安装脚本

官方的脚本太慢，国内用户使用这个脚本进行加速

$ curl -sfL https://rancher-mirror.rancher.cn/k3s/k3s-install.sh | INSTALL_K3S_MIRROR=cn sh -

检查k3s服务状态

$ systemctl status k3s

查看集群节点状态

$ sudo kubectl get node

查看集群中运行的pod

$ sudo kubectl get pods -A

K3s 官方经典架构与安装文档

二、配置kubectl （普通用户）

普通用户执行命令 kubectl get nodes 提示没有权限

pig@vm-debian:~$ mkdir -p /.kube #创建 .kube 目录 pig@vm-debian:$ sudo cp /etc/rancher/k3s/k3s.yaml /.kube/config #复制 K3s 默认配置 pig@vm-debian:$ sudo chown $USER:$USER /.kube/config #修改文件所属用户 pig@vm-debian:$ echo ’export KUBECONFIG=$HOME/.kube/config’ >> ~/.bashrc #配置环境变量 source /.bashrc pig@vm-debian:$ kubectl get nodes #验证配置 NAME STATUS ROLES AGE VERSION vm-debian-13 Ready control-plane 175m v1.35.5+k3s1

如上，说明当前普通用户已经可以正常访问 Kubernetes API。

运维思考

生产环境中，不建议长期使用root用户进行Kubernetes管理。

推荐使用普通用户，通过 sudo 执行系统级操作（如管理服务、修改配置），而将 kubectl 配置为普通用户直接使用。

这样既符合Linux最小权限原则，也更符合企业运维的实际工作方式。

三、配置dashboard

为了避免下载失败的情况，这里使用国内可访问的镜像仓库

3.1下载YAML

wget https://raw.githubusercontent.com/kubernetes/dashboard/v2.7.0/aio/deploy/recommended.yaml

3.2修改类似内容

找到类似内容：


image: kubernetesui/dashboard:v2.7.0
改成（示例）：
image: docker.m.daocloud.io/kubernetesui/dashboard:v2.7.0
再找到：
image: kubernetesui/metrics-scraper:v1.0.8
改成：

image: docker.m.daocloud.io/kubernetesui/metrics-scraper:v1.0.8

3.3开始部署

$ kubectl apply -f recommended.yaml namespace/kubernetes-dashboard unchanged serviceaccount/kubernetes-dashboard unchanged service/kubernetes-dashboard unchanged secret/kubernetes-dashboard-certs unchanged secret/kubernetes-dashboard-csrf unchanged secret/kubernetes-dashboard-key-holder unchanged configmap/kubernetes-dashboard-settings unchanged role.rbac.authorization.k8s.io/kubernetes-dashboard unchanged clusterrole.rbac.authorization.k8s.io/kubernetes-dashboard unchanged rolebinding.rbac.authorization.k8s.io/kubernetes-dashboard unchanged clusterrolebinding.rbac.authorization.k8s.io/kubernetes-dashboard unchanged deployment.apps/kubernetes-dashboard configured service/dashboard-metrics-scraper unchanged deployment.apps/dashboard-metrics-scraper configured

3.4验证是否成功

这里有2个之前创建创建的pod

$ kubectl get pods -n kubernetes-dashboard NAME READY STATUS RESTARTS AGE dashboard-metrics-scraper-7764d6f57c-r26sm 0/1 ContainerCreating 0 13m dashboard-metrics-scraper-8d46b45f6-4c6bv 0/1 ContainerCreating 0 23h kubernetes-dashboard-659cf4fdd5-l2gp9 0/1 ContainerCreating 0 13m kubernetes-dashboard-b44857bbb-9k9kc 0/1 ContainerCreating 0 23h

这里READY一直是0，失败了，需要排查一下。

查看pod卡在哪里了

kubectkubectl describe pod -n kubernetes-dashboard-659cf4fdd5-l2gp9 No resources found in kubernetes-dashboard-659cf4fdd5-l2gp9 namespace. pig@vm-debian:$ pig@vm-debian:$ kubectl describe pod -n kubernetes-dashboard dashboard-metrics-scraper-7764d6f57c-r26sm |tail Warning FailedCreatePodSandBox 17m kubelet Failed to create pod sandbox: rpc error: code = DeadlineExceeded desc = failed to start sandbox “74e4795e553dc82c49e53c89a724293525e65e3134fa2ae09787ff197831d938”: failed to get sandbox image “rancher/mirrored-pause:3.6”: failed to pull image “rancher/mirrored-pause:3.6”: failed to pull and unpack image “docker.io/rancher/mirrored-pause:3.6”: failed to resolve reference “docker.io/rancher/mirrored-pause:3.6”: failed to do request: Head “https://registry-1.docker.io/v2/rancher/mirrored-pause/manifests/3.6": dial tcp 182.50.139.56:443: i/o timeout-n kubernetes-dashboard-659cf4fdd5-l2gp9 No resources found in kubernetes-dashboard-659cf4fdd5-l2gp9 namespace. pig@vm-debian:$ pig@vm-debian:$ kubectl describe pod -n kubernetes-dashboard dashboard-metrics-scraper-7764d6f57c-r26sm |tail Warning FailedCreatePodSandBox 17m kubelet Failed to create pod sandbox: rpc error: code = DeadlineExceeded desc = failed to start sandbox “74e4795e553dc82c49e53c89a724293525e65e3134fa2ae09787ff197831d938”: failed to get sandbox image “rancher/mirrored-pause:3.6”: failed to pull image “rancher/mirrored-pause:3.6”: failed to pull and unpack image “docker.io/rancher/mirrored-pause:3.6”: failed to resolve reference “docker.io/rancher/mirrored-pause:3.6”: failed to do request: Head “https://registry-1.docker.io/v2/rancher/mirrored-pause/manifests/3.6": dial tcp 182.50.139.56:443: i/o timeout

Kubernetes 无法拉取 pause 镜像（sandbox基础镜像）

解决方案：


#修改 K3s 的 pause 镜像
$ sudo vi  /etc/rancher/k3s/registries.yaml #编辑以下文件
mirrors:
  docker.io:
    endpoint:
      - "https://docker.m.daocloud.io"
#然后进行重启
$sudo systemctl restart k3s

$ kubectl get pods -n kubernetes-dashboard #这个时候已经创建成功了 NAME READY STATUS RESTARTS AGE dashboard-metrics-scraper-7764d6f57c-r26sm 1/1 Running 0 28m kubernetes-dashboard-659cf4fdd5-l2gp9 1/1 Running 1 (2m45s ago) 28m

3.4配置Dashboard进行访问

可以通过 kubectl proxy，或者配置NodePort进行进行访问。

我们这里通过配置NodePort来继续。

$kubectl edit svc -n kubernetes-dashboard kubernetes-dashboard #找到： type: ClusterIP #修改成 type: NodePort

$kubectl get svc -n kubernetes-dashboard #查看端口

主机访问：https://ip:port

生成token

$ kubectl -n kubernetes-dashboard create token admin-user eyJhbGciOiJSUz…..qGeyA #这里进行省略输出

获取token并进行登录

至此，K3s 单节点环境已完成基础验证，包括：

kubectl 可用
Pod 正常运行
Dashboard 可访问并登录成功