This message was deleted.

That is not supported. The downstream cluster needs to be able to reach rancher. That is how the agent is designed - once rancher is deployed, all communication is initiated by agents in the downstream cluster. It will not ever function if your environment does not allow this.

Oh - OK. I thought there was a "poll" management method from Rancher -> managed cluster.

Nope. Cluster agent opens a connection to rancher and everything happens via that.

OK, thanks for the assist!