The problem that machine nodes cannot be reused, a node cannot be both a master and a worker

[vpc123]

Issue Description

Type: feature request

Describe what feature you want

操作系统: centos7.x 安装命令: sealer run kubernetes:v1.19.8 -m 192.168.139.140 -n 192.168.139.140 -p 12589

报错原因: 之前遗留的配置资源干扰集群安装,根本原因是因为，安装命令将节点既指定为 master 同时也指定为 worker ，master 已经将资源下发，所以 worker 节点时发生了资源冲突；

说明: sealer 安装集群时无法使得节点复用，节点复用作为一种可选的功能对用户暴露可以增强 sealer 的扩展性；

Additional context

Add any other context or screenshots about the feature request here.

[allencloud]

@vpc123 Thanks greatly for your report. I think actually we should show more flag definition in the sealer run --help command. When you input the --master IP_A, IP_A will be set the role of MASTER, and it is also a WORKER role, indeed, in the scope of Kubernetes.

So, it may break someone's original mind about Kubernetes knowledge.

Back to your experiment, you only need to execute sealer run sealer run kubernetes:v1.19.8 -m 192.168.139.140 -p 12589, and 192.168.139.140 is both master and worker.

For the sealer's design on master and node definition, could you leave your suggestion how to better it here?

[kakaZhou719]

@allencloud @vpc123 ,we could pre valid the master and node ip, if they are the same one, return command input error? Or just go ahead install clusterImage on one node?