You most likely execute nodes 1 by 1 following the flow.
You need to run the entire workflow. However it is not recommended as the loop is endless and will eventually consume memory and crash the server.
Consider adding an If node with a condition to terminate the loop at some iteration value before testing the entire flow in a single full execution.
If this resolves your question, please mark this post as a Solution.
But why it doesn’t work if I run manually one node at a time? This is the whole point of doing that BTW, to check how the process goes on in a “debug mode” (sort of)
You can run one node at a time, but you need to pin data to the previous node.
SInce the workflow is all meant to work together, each node relies on the one before it for the input data. If you go in and click the pencil icon you can edit mock data to save to do that
I don’t understand: if a node has the possibility to have two nodes at its input, why when running everything manually step-by-step it ignores the latest input data?
If you want the assurance that a node won’t run again until you want it to, if it just ran you can press P on the keyboard and that’s the shortcut to pin the data that’s currently in it