I think the error results from different networks of the two nodes(one 192.168.1.57 and the other 192.168.60.67). So I use two other machines with IP addresses 192.168.206.100(as a worker) and 192.168.206.101(as a params server) respectively. Similarly I use a docker container in each machine.
It successfully solves the above mismatch error, but a new error occurs on the worker node: “… connection to [172.17.0.10]:5807 is refused”. “172.17.0.10” is the IP address of the container and “5807” is a random port. Notice that the master address I set is 192.168.206.101 and the master port 60000 as above.
In the disscussion, @lcw says some “real” connections exist. The random port is created by one of these connections.
###########
##########
As only the port 60000 of the container is mapped to the same port of the host(server), a random port(such as 5807) may not be accessed. So the connection to such a random port is refused.
Can I still use a docker container on each machine?
thanks.