My current cluster environment is as follows:

~ $ docker node ls
 ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUS
 czn1ts6u2o24pcyeojlsr9ai7     node4               Ready               Active
 lrlhx7smzo9rafckep5b05fgw     node3               Ready               Active
 s38rdi1wlxqbvtfl1sa3ne17r *   node1               Ready               Active              Leader
 uubznl6kas1mpd5evpqamtr9x     node5               Ready               Active
 z4otpoy5a8m8px2e3bh9jy5jd     node2               Ready               Active

Then i promote node2, node3.

~ $ docker node promote node2 node3
 Node node2 promoted to a manager in the swarm.
 Node node3 promoted to a manager in the swarm.

Check the current node status again.

~ $ docker node ls
 ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUS
 czn1ts6u2o24pcyeojlsr9ai7     node4               Ready               Active
 lrlhx7smzo9rafckep5b05fgw     node3               Ready               Active              Reachable
 s38rdi1wlxqbvtfl1sa3ne17r *   node1               Ready               Active              Leader
 uubznl6kas1mpd5evpqamtr9x     node5               Ready               Active
 z4otpoy5a8m8px2e3bh9jy5jd     node2               Ready               Active              Reachable

Then I started to STOP leader node (manager)

~ $ docker-machine stop node1
 Stopping "node1"  ...
 Machine "node1" was stopped.

Execute the following command again

~ $ docker node ls
 error during connect: Get https://192.168.99.100:2376/v1.30/nodes: dial tcp 192.168.99.100:2376: getsockopt: no route to host

Then the question arises, how do I know who is the leader node? , because I only know leader node, began to carry out

eval $(docker-machine env 'leader node name')