Docker Version Rancher Cluster Regularly Restart (Even Shutdown)

I start a rancher cluster by start a rancher server in docker mode. I know this is not the HA mode for production environment. But I still think what I get is quite unreasonable.

On the last weekend my cluster shutdown and I got to know about this in the morning. When I login the rancher server machine I found that all the containers are stopped. But after a while all the service start gradually…So it seems like all the services only restart after I login in the machine…They have already stopped more than 24 hours before that. And this is not the first time I saw this issue. But this time is do shutdown for longest time.

BTW my cluster is created in a very heterogeneous network: The master and two agents are in Singapore, one agent is in Beijing and other slave is in Tokyo. Does this is related with the shutdown?