Coordinated approaches between datacenter and non-datacenter loads in buildings to achieve energy efficiency have been widely studied in recent years. However, the coordination-enabled mechanisms still leave much room to be optimized for the following reasons. First, in the new trend of cloud computing networks, datacenters are pushed to the edge of networks to reduce latency (they are called edge datacenters). Such datacenters are often deployed in geographically distributed buildings and collocated with offices in terms of sharing building infrastructure; such buildings are called geo-distributed mixed-use buildings (MUBs). That scenario has not been well addressed in terms of energy sustainability requirements overall in the buildings; the requirements are imposed either by mandatory government orders or by LEED certification. Second, one critical issue of datacenters is water saving, which is rarely associated with energy efficiency, even though every kilowatt of energy consumption can reflect exactly an amount of water use in datacenters. Therefore, in this paper, we aim to find a solution for joint energy scheduling and water saving problem ( PJEW ) in a coordinated manner between MUBs. The solution is designed to schedule workloads by coupling edge datacenters collocated in buildings as well as to control energy and water usage to minimize the system cost caused by reducing loads. We advocate the model predictive control to schedule the whole system in a time horizon. Multiple simulation scenarios are evaluated to show the efficiency of our proposed methods compared to conventional approaches. The results reveal that our mechanism outperforms the uncoordinated methods and achieves a nearly optimal solution.