Understand the nature of the gravity is one of the most important motivations of future large galaxy surveys such as Euclid and DESI. Although the accuracy of the measurements of galaxy clustering has been improved dramatically during the past decade, approaching percent level, there are still some challenges to make a final conclusion of theory of gravity and tell the fate of the Universe. This is largely due to the potential systematics in observations and uncertainties in theories.  In this talk, I will talk about major challenges and systematics in testing gravity on cosmological scales and also discuss possible ways to get around them. My talk has three parts. In the first part, I will talk about the challengeson the theory side. In the second part, I will talk about systematics in real data and how to mitigate them. In the last part, I will compare theory with data and show that predictions in LCDM match data very well. Therefore, the possibilities of modified theories of gravity are diminished.