Voltage/Var optimization (VVO) is an important operational practice for distribution networks. In recent years, the distribution networks have been evolving towards more active systems, called active distribution networks (ADNs). In particular, as a power-electronic device, the solar PV inverters have a very fast-acting and flexible reactive power support capability. Therefore, they can also be employed for VVO purposes. On the other hand, the solar PV has a highly random power generation characteristic, which would make the bus voltage fluctuation also highly random and may even violate the voltage magnitude constraints. In order to address such random voltage fluctuation problem, stochastic VVO is needed. In this paper, a stochastic optimization based VVO approach is proposed for ADNs with customer-owned solar PV units. The PV inverters are utilized to provide reactive power support in addition to the OLTC and capacitor banks. A sampling average approximation (SAA) method is also proposed to solve the stochastic optimization problem. The proposed approach is tested on the IEEE 33-bus system, and the simulation results validate its effectiveness and advantages over the traditional deterministic VVO method.