I have the following setup: behind a NAT, I have two machines, let’s call them the big machine and the small machine, both running ubuntu, and I can access both with the same username and each can ssh into the other. Both also have ray installed.
On the large machine, I run the coordinator_server.py script as recommended in the docs and using the ip of the small machine as the
I then create a config based on the minimal_automatic one, with the coordinator host and port from the previous step.
Next, i run
ray up -y my_automatic_config.yaml, which eventually says that ray runtime is started.
Once all that is done, I run the testing script (from /cluster/quickstart.html in the docs) to check that all the nodes are available.
But it only shows the cluster consisting of one node, the large machine.
What am I doing wrong? Is there something I’ve missed in the docs, the tutorials, setting up the cluster, or anything else?