Your issue is a result of making calls to the Revit API, causing nodes to fail very early in the graph (and repeatedly up stream from that).
Open the graph in Dynamo Sandbox (in program files → Dynamo) instead of Dynamo for Revit, and you’ll see that your second python node fails, as does EVERY node after that. This is because in Dynamo Sandbox there is no connection to the Revit API as it is not Dynamo for Revit.
This Dynamo Sandbox environment is similar to how Refinery runs, with no connection to any Revit environment. The Remember nodes will allow you to pull data into the Sandbox environment (and Refinery), but not facilitate direct Revit interaction they way your graph is attempting. In Dynamo for Revit, remove all python code which calls the Revit API, and set up Remember nodes to work with the data queried from the Revit model. Then save the graph, close the Revit model (and Revit - your CPU will want the resources) and open Dynamo Sandbox. Then open your graph and run Refinery from there.