I am trying to transfer a large directory from a pod on a downstream cluster via rancher to a local directory, and it is failing with unanswered websocket pings. Looking through wireshark on the node running the rancher pod it is talking to, basically shows that the TCP window fills and no response to the pings, and kubectl fails after a minute of failed pings, smaller transfers succeed ok, this is apparent only with a recent upgrade, which has unfortunately changed a few things at once. Do you have any suggestions on how to determine where in the path the problems are occuring?

It may not be this, but my immediate thought is that the downstream cluster node has DCHP and the IP changed. Setting a reservation would fix this.

Direct to the downstream cluster works fine. I guess I need to try to test the remaining links thanks