Due to the continuous increase in data traffic, communication networks require increased densification and multi-layering. Different hotspot areas may access different networks. We aim to avoid interference from cell-free millimeter-wave (mmWave) massive multiple-input multiple-output (MIMO) sys-tems in hotspot areas. To this end, we consider a specific scenario involving the cell-free mmWave massive MIMO system and the mmWave Vehicle-to-Everything (V2X) network. To minimize interference on the mmWave V2X network while ensuring the sum spectral efficiency (SE) performance of the cell-free mmWave massive MIMO system, a novel joint problem of beam selection and user equipment (UE) association is formulated. We propose a deep reinforcement learning (DRL)-based distributed double deep Q-network (D-DDQN) algorithm to address this problem. Simulation results indicate that the algorithm we propose out-performs traditional algorithms.