In this letter, we investigate the power minimization problem in an Internet of Things (IoT) sensor network assisted by reconfigurable intelligent surface (RIS). Different from the existing literature, we consider both transmit and non-transmit energy consumption, which leads to a joint sensor selection, beamforming design and RIS phase configuration problem. Based on the cutting-the-edge majorization minimization (MM) and penalty duality decomposition (PDD) frameworks, we develop two solutions which can achieve nearly identical performance as that of exhaustive search method. Extensive numerical results verify the great benefit in power saving via sensor selection and the deployment of RIS device.