k3s集群创建及导入rancher

卸载K3S
/usr/local/bin/k3s-uninstall.sh

/usr/local/bin/k3s-agent-uninstall.sh

所有步骤具体步骤如下:
准备两台服务器，一台做master，一台做work

#设置Hostname
hostnamectl set-hostname $hostname

#添加hosts记录
echo "$host_ip $hostname" >> /etc/hosts

#关闭Selinux
sudo sed -i 's/SELINUX=enforcing/SELINUX=disabled/g' /etc/selinux/config
setenforce 0

#关闭防火墙
systemctl stop firewalld.service && systemctl disable firewalld.service

#检查与设置时区与时间
timedatectl status

#设置时区
ln -sf /usr/share/zoneinfo/Asia/Shanghai /etc/localtime
timedatectl set-timezone Asia/Shanghai

安装k3s

master 节点，安装k3s

curl -sfL http://rancher-mirror.cnrancher.com/k3s/k3s-install.sh | INSTALL_K3S_MIRROR=cn INSTALL_K3S_VERSION=v1.20.8+k3s1 sh - ​

node 节点安装k3sk

1. 从master获取K3S_URL和K3S_TOKEN
   cat /var/lib/rancher/k3s/server/node-token

2. 复制粘贴成如下形式，得到K3S_TOKEN
   k3s_token=“K1055e0336c8ef095770165a1b9ab7db35b73f352109916438e1c85aaf9abe73358::server:28f7682fb16e799ea283f7c2b46767f0”

3. 查看master主机IP，（注意：master主机和work主机可以通信。）命令如下：
   k3s_url=“https://ip:6443”

4. 在work主机执行
   curl -sfL http://rancher-mirror.cnrancher.com/k3s/k3s-install.sh | INSTALL_K3S_MIRROR=cn INSTALL_K3S_VERSION=v1.20.8+k3s1 K3S_URL=${k3s_url} K3S_TOKEN=${k3s_token} sh -
   ​

5. 执行命令检查是否成功：k3s kubectl get node
   ​
   “”"
   [root@k3s-master server]# kubectl get nodes
   NAME STATUS ROLES AGE VERSION
   k3s-master Ready control-plane,master 3m30s v1.21.1+k3s1
   k3s-node01 Ready 3s v1.21.1+k3s1
   “”"

将k3s集群导入rancher中

1. 直接在rancher-ui添加集群，选择导入，集群名字自定义

2. 确定后，因为这边使用的rancher自带证书，所以复制第三个命令到k3s集群master主机的终端运行
   curl --insecure -sfL https://rancher.skybility.org/v3/import/htzkpn2576pj56dwl8jhc4wg96547rknc4gf4w5jlpprg8578sxqfm_c-9jxmd.yaml | kubectl apply -f -​

3.导入成功，集群状态为active

导入k3s集群出现的问题

1.集群添加不成功，一直处于pending，发现问题如下

[root@k3s-master /]# kubectl get pods --all-namespaces
NAMESPACE       NAME                                     READY   STATUS             RESTARTS   AGE
kube-system     metrics-server-86cbb8457f-54vds           1/1     Running           0         66m
kube-system     local-path-provisioner-5ff76fc89d-cf7hd   1/1     Running           0         66m
kube-system     coredns-7448499f4d-tcjpd                 1/1     Running           0         66m
kube-system     helm-install-traefik-crd-s9kfc           0/1     Completed         0         66m
kube-system     helm-install-traefik-w84qr               0/1     Completed         1         66m
kube-system     svclb-traefik-ts9ww                       2/2     Running           0         65m
kube-system     traefik-97b44b794-fz6wz                   1/1     Running           0         65m
kube-system     svclb-traefik-8hfv5                       2/2     Running           0         63m
cattle-system   cattle-cluster-agent-7487885496-zm2xr     0/1     CrashLoopBackOff   3         114s

​
2.查看日日志

[root@k3s-master /]# kubectl logs cattle-cluster-agent-7487885496-zm2xr -n cattle-system
INFO: Environment: CATTLE_ADDRESS=10.42.0.9 CATTLE_CA_CHECKSUM=646733538a1a16126ff41d170a04d1791fa591fe38ff1b292e02ea54b124e17e CATTLE_CLUSTER=true CATTLE_CLUSTER_REGISTRY=registry.cn-hangzhou.aliyuncs.com CATTLE_FEATURES= CATTLE_INTERNAL_ADDRESS= CATTLE_IS_RKE=false CATTLE_K8S_MANAGED=true CATTLE_NODE_NAME=cattle-cluster-agent-7487885496-zm2xr CATTLE_SERVER=https://rancher.skybility.org
INFO: Using resolv.conf: search cattle-system.svc.cluster.local svc.cluster.local cluster.local nameserver 10.43.0.10 options ndots:5
ERROR: https://rancher.skybility.org/ping is not accessible (Could not resolve host: rancher.skybility.org)

​
问题很明显，虽然客户端配置了 hosts，但是集群节点并不知道 rancher.skybility.org 是哪个机器。
​
3.解决方法
​

在所有节点配置 /etc/hosts，先让所有节点都能认识 rancher.skybility.org

vim  /etc/hosts
# 添加如下配置
ip rancher.skybility.org

为Agent Pod添加主机别名(/etc/hosts)

kubectl -n cattle-system patch deployments cattle-cluster-agent --patch '{
  "spec": {
      "template": {
          "spec": {
              "hostAliases": [
                  {
                      "hostnames":
                      [
                          "ns.skyrancher.com"
                      ],
                          "ip": "192.168.4.96"
                  }
              ]
          }
      }
  }
}'

​

如果需要切换容器引擎,默认为container

# 修该容器引擎，定义变量 INSTALL_K3S_EXEC
export INSTALL_K3S_EXEC="--docker --write-kubeconfig ~/.kube/config --write-kubeconfig-mode 666"
curl -sfL https://docs.rancher.cn/k3s/k3s-install.sh | INSTALL_K3S_MIRROR=cn sh -
​
命令参数解释：
INSTALL_K3S_EXEC="（略）"效果为启动k3s服务时使用的额外参数。
--docker ：效果为使用docker而不是默认的containerd。
--write-kubeconfig-mode 666 ：效果为将配置文件权限改为非所有者也可读可写，进而使kubectl命令无需root或sudo。
--write-kubeconfig ~/.kube/config ：效果为将配置文件写到k8s默认会用的位置，而不是k3s默认的位置/etc/rancher/k3s/k3s.yaml。后者会导致istio、helm需要额外设置或无法运行。

问题二

1. 导入完成后，一直处于wait状态，查看日志，出现以下错误，dns问题

time="2021-06-21T01:12:44Z" level=error msg="failed to call leader func: settings.management.cattle.io \"install-uuid\" not found"
time="2021-06-21T01:12:49Z" level=error msg="failed to call leader func: settings.management.cattle.io \"install-uuid\" not found"