This paper suggests a novel approach to the problem of online optimal control of reactive sources in wind power plants (WPPs). Based on measurements tracking the actual conditions of each wind generator and the settings of transformers and compensation equipment within the WPP facility, the key idea is to incorporate reactive power optimization into a global WPP's control loop so that it can be used online to determine the optimal distribution of reactive power, which is needed to meet grid code requirements, among the available Var sources. The optimization problem is handled using an enhanced version of the mean-variance mapping optimization algorithm. A test case, based on a representative offshore WPP, is presented to demonstrate the effectiveness of the proposed control strategy.