1.安装前环境清理：
找到安装时的节点，进入kubeasz容器，进入/etc/ansible目录：
查看docker运行的项目：docker ps
使用名称或id进入kubeasz的容器：docker exec -it a5f466748bc6 bash
进入/etc/ansible目录：cd /etc/ansible/
bash-4.4# ls
01.prepare.yml 04.kube-master.yml 11.harbor.yml 90.setup.yml README.md bin example manifests tools
02.etcd.yml 05.kube-node.yml 22.upgrade.yml 91.start.yml a dockerfiles gcluster-service.yaml pics
03.containerd.yml 06.network.yml 23.backup.yml 92.stop.yml ansible.cfg docs hosts pssh
03.docker.yml 07.cluster-addon.yml 24.restore.yml 99.clean.yml ansible.cfg.rpmnew down hosts.rpmnew roles
执行清理命令：
bash-4.4# ansible-playbook 99.clean.yml
卸载后检查：
检查各节点etcd.service、kube-apiserver、kubelet是否都卸载干净(例：systemctl status etcd)
卸载后建议重启一下节点。
###建议找几个环境比较干净的节点去搭建，能少解决非常多问题
#安装节点可以独立于k8s集群，也可以是k8s集群中的节点
2.安装前准备-->配置k8s集群各节点间互信、安装节点和k8s集群节点互信
配置互信命令：ssh-copy-id <ip>
遇到的问题：
①[root@songyanling-node-3 ~]# ssh-copy-id 10.10.11.16
/usr/bin/ssh-copy-id: ERROR: failed to open ID file '/root/.pub': No such file or directory
(to install the contents of '/root/.pub' anyway, look at the -f option)
解决方法：ssh-keygen -t dsa，回车回车回车，然后再次配置互信
②[root@jubin-node-1-0 ~]# ssh-copy-id 10.10.3.74
/usr/bin/ssh-copy-id: INFO: Source of key(s) to be installed: "/root/.ssh/id_rsa.pub"
/usr/bin/ssh-copy-id: INFO: attempting to log in with the new key(s), to filter out any that are already installed
/usr/bin/ssh-copy-id: ERROR: @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
ERROR: @ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @
ERROR: @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
ERROR: IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!
ERROR: Someone could be eavesdropping on you right now (man-in-the-middle attack)!
ERROR: It is also possible that a host key has just been changed.
ERROR: The fingerprint for the ECDSA key sent by the remote host is
ERROR: SHA256:EQzIyLTcC1ZLWi+c9Kf+sLD7G9yQJhwiImwebLs6C28.
ERROR: Please contact your system administrator.
ERROR: Add correct host key in /root/.ssh/known_hosts to get rid of this message.
ERROR: Offending ECDSA key in /root/.ssh/known_hosts:21
ERROR: ECDSA host key for 10.10.3.74 has changed and you have requested strict checking.
ERROR: Host key verification failed.
解决方法：在执行节点上删除/root/.ssh/known_hosts后，再次执行互信
3.安装-->在安装节点上安装docker、安装ansible并启动
①安装并启动docker：
在10.10.55.76节点上有docker安装包：/opt/docker-ce-1703（建议使用这个docker包，因为yum install docker安出来的可能有版本问题）
拷贝到自己的节点，然后执行安装，安装顺序如下：
rpm -ivh docker-ce-selinux-17.03.2.ce-1.el7.centos.noarch.rpm
rpm -ivh docker-ce-17.03.2.ce-1.el7.centos.x86_64.rpm
（如果安装报错缺少包的话，百度一下把缺少的包安上就行）
启动docker：systemctl start docker
②安装并启动ansible
在10.10.11.2的/etc目录下有ansible.tar，拷贝到安装节点的/etc下解压：tar -xvf xxx.tar
启动ansible：进入 /etc/ansible/tools，执行：./easzup -S
进入kubeasz容器：docker exec -it kubeasz bash
然后到/etc/ansible目录下：先配置hosts
配置完成后依次执行以下.yml的安装，例如：ansible-playbook 01.prepare.yml
01.prepare.yml 04.kube-master.yml
02.etcd.yml 05.kube-node.yml
06.network.yml
03.docker.yml 07.cluster-addon.yml
4.安装后检查
kubectl get node查看所有node和master节点是否都是ready状态
如果有节点是not ready，到节点上看是否是kubelet服务有问题，如果有问题，查看free -g看swap是否都为0，如果不是执行swapoff -a
5.安装后推镜像
一般选取k8s的master节点推镜像就行
(1)解压集群安装包后，进入gcdw/image目录下，有8个镜像
①docker load -i <镜像.tar>
例：docker load -i server-9.9.0.7.4-202112161240.tar
20c8fba63ae3: Loading layer [==================================================>] 241.1MB/241.1MB
5c0b176700ce: Loading layer [==================================================>] 289.1M192.168.8.84/gcdw/gcdw-server:9.8.0B/289.1MB
Loaded image: 192.168.8.84/gcdw/gcdw-server:9.9.0.7.4
②docker tag 192.168.8.84/gcdw/gcdw-server:9.9.0.7.4 harbor.gbase.cn/gcdw/gcdw-server:9.9.0.7.4
③docker push harbor.gbase.cn/gcdw/gcdw-server:9.9.0.7.4
(2)推镜像可能遇到的报错
①报错一
docker push harbor.gbase.cn/gcdw/gcdw-operator:9.8.0.4.1
The push refers to a repository [harbor.gbase.cn/gcdw/gcdw-operator]
Get https://harbor.gbase.cn/v1/_ping: dial tcp: lookup harbor.gbase.cn on [2001:db8a:8620:9530::2]:53: dial udp [2001:db8a:8620:9530::2]:53: connect: network is unreachable
解决方法：echo '10.10.1.1 harbor.gbase.cn' >> /etc/hosts
②报错二
docker push harbor.gbase.cn/gcdw/gcdw-operator:9.8.0.4.1
The push refers to a repository [harbor.gbase.cn/gcdw/gcdw-operator]
Get https://harbor.gbase.cn/v1/_ping: dial tcp 10.10.1.1:443: getsockopt: connection refused
解决方法：
cat > /etc/docker/daemon.json << EOF
{
"registry-mirrors": ["http://harbor.gbase.cn"],
"insecure-registries": ["harbor.gbase.cn"]
}
EOF
修改后重新加载daemon，并重启docker: systemctl daemon-reload;systemctl restart docker
③报错三
docker push harbor.gbase.cn/gcdw/gcdw-operator:9.8.0.3.7
The push refers to a repository [harbor.gbase.cn/gcdw/gcdw-operator]
9cec9d938f49: Preparing
4bcb77274d4a: Preparing
071d8bd76517: Preparing
denied: requested access to the resource is denied
解决方法：需要docker login harbor.gbase.cn 用户名密码为admin/Harbor12345
