I don’t think you can easily formulate this problem with SDDP.jl. It doesn’t support chance constraints.
What is the interpretation of the constraint? Why is it necessary?
If you were to formulate it, you would need to add \eta_v as a state variable, and then you must decide the value of \eta_v in node n. But SDDP.jl isn’t really designed to handle this case because it requires the state dimension to be constant across nodes, so you’ll probably run into a bunch of issues.