In this paper, we investigate a MIMO relaying network with massive antennas for serving multi-pair single-antenna user s, where zero-forcing (ZF) precoding is explored at the relay Our aim is to maximize the energy efficiency (EE) by jointly optimizing the transmit power at the source-nodes and the relay, and the number of relay antennas. To solve the EE optimization problem, a lower bound of the involved spectra efficiency (SE) is derived with analytical expression. Since the original fractional optimization problem is non-convex we transform the objective function into an equivalent parametric subtractive form by using the properties of fractional programming. Then, an iterative algorithm is developed relying on the partial convexity of the subtractive cost function where the analytical solutions are deduced. Finally, the effectiveness and convergence behavior of our proposed schemi are verified through the numerical results.