1.了解Raft协议

Raft协议：保证大多数节点存活才可用。只要>1 ，集群至少大于3台，至少存在3个master节点，如果一个master节点挂了，则会从另外两个备节点选举。

2.具体实验操作

操作1，将主节点服务停止
1、之前Docker swarm 集群部署搭建了集群，而且有一个备master节点，现在将leader节点干掉
干掉前

[root@l4 ~]# docker node ls
ID                            HOSTNAME   STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
kpsnt4qat4ib6mc7axkm1fkok     l4         Ready     Active         Leader           20.10.9
sf5lcgbh6329zoogsfev4kp5d     l4         Ready     Active                          20.10.9
v566v6m67v3dpqrmciyy6ejyv     l4         Ready     Active                          20.10.9
xq4ojhweyv5htnx88x5l9wvmc *   l4         Ready     Active         Reachable        20.10.9

[root@l4 ~]# systemctl stop docker
Warning: Stopping docker.service, but it can still be activated by:
  docker.socket

发现其他主机获取不到节点信息，说明另外一个主节点不能使用了

[root@l4 ~]# docker node ls
Error response from daemon: rpc error: code = DeadlineExceeded desc = context deadline exceeded

重新启动之前干掉的master节点

[root@l4 ~]# systemctl start docker

再次获取node节点信息，发现之前的备主节点已经是主节点了

操作2，将worker节点停掉，变成master节点
1.将10.7移除集群

[root@l4 ~]# docker swarm leave
Node left the swarm.

发现10.7主机状态已经是Down

[root@l4 ~]# docker node ls
ID                            HOSTNAME   STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
kpsnt4qat4ib6mc7axkm1fkok     l4         Ready     Active         Reachable        20.10.9
sf5lcgbh6329zoogsfev4kp5d     l4         Down      Active                          20.10.9
v566v6m67v3dpqrmciyy6ejyv     l4         Ready     Active                          20.10.9
xq4ojhweyv5htnx88x5l9wvmc *   l4         Ready     Active         Leader           20.10.9

2.之后将10.7变成备主节点
主节点和备主节点都可以执行获取master令牌命令，worker节点不行

[root@l4 ~]# docker swarm join-token manager
Error response from daemon: This node is not a swarm manager. Use "docker swarm init" or "docker swarm join" to connect this node to swarm and try again.
[root@l4 ~]# docker swarm join --token SWMTKN-1-0o6vjc79zgadg3aw0a18ygeway7v5db21cp74q1dea3gkgl49z-bx0smnmm22qfbf5g0p05k8g9u 192.168.10.6:2377
This node joined a swarm as a manager.

3.现在集群节点状态，有三个节点是主节点（1主，两备）

[root@l4 ~]# docker node ls
ID                            HOSTNAME   STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
kpsnt4qat4ib6mc7axkm1fkok     l4         Ready     Active         Reachable        20.10.9
lqu6ybb2l3sals78rtgv0tv0t *   l4         Ready     Active         Reachable        20.10.9
sf5lcgbh6329zoogsfev4kp5d     l4         Down      Active                          20.10.9
v566v6m67v3dpqrmciyy6ejyv     l4         Ready     Active                          20.10.9
xq4ojhweyv5htnx88x5l9wvmc     l4         Ready     Active         Leader           20.10.9

4.将主节点10.6 docker服务停掉，查看

[root@l4 ~]# systemctl stop docker
Warning: Stopping docker.service, but it can still be activated by:
  docker.socket

之后查看节点信息，发现可以查看到，而且主节点已经变成其他主机了

[root@l4 ~]# docker node ls
ID                            HOSTNAME   STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
kpsnt4qat4ib6mc7axkm1fkok     l4         Ready     Active         Reachable        20.10.9
lqu6ybb2l3sals78rtgv0tv0t *   l4         Ready     Active         Leader           20.10.9
sf5lcgbh6329zoogsfev4kp5d     l4         Down      Active                          20.10.9
v566v6m67v3dpqrmciyy6ejyv     l4         Ready     Active                          20.10.9
xq4ojhweyv5htnx88x5l9wvmc     l4         Unknown   Active         Unreachable      20.10.9

再次干掉leader节点 则不能再被使用