Hi folks, having a quick question about how rancher server communicates with cattle-agent. As it's an external container hosted on a docker separated machine, it cannot contact the Launched Kubernetes Cluster internals IPs, am i right ? So it needs some kind of port forwarding or smtg ? Thanks in advance

When a downstream cluster is set up in Rancher, a kubectl command is run on the downstream cluster which installs the cluster agent. This same command gives the cluster agent a token. The agent then initiates communication with Rancher which is authenticated with that token. I'm not sure if that answers your question

Thanks for your reply 🙂 . My question was more related to how the local cluster communicates with the downstream one ? I saw many errors on the cluster explorer where the healthcheck could not be checked on the downstream cluster. The cluster API endpoint that been queried was a special port (46634/TCP). But as i read the documentation you provided, it seems that the only port needed is the 6443/TCP.

The Kubernetes API server runs on 6443. I have no idea where 46634 came from

Okay, thanks anyway. I was just wondering about this error. Restarting the rancher container solved the problem.