Sorry, my mistake, I misread the problem. You are perfectly right.
Thing that you have to change is to calculate concentration of CO32- from the presence of sodium carbonate (using similar Ka2 based approach you already used). Just add this concentration to the CO32- concentration from MgCO3 dissolution in the final step, where you plug the [Mg2+] and [CO32-] into Ksp.
And I hope I am right this time.