Special characteristics of millimeter wave (mmWave) systems such as high vulnerability to random obstacles (due to high penetration loss) and mobility (due to directional communications) demand redesigning the existing algorithms or the association between clients and access points. In this paper, we propose a novel dynamic association scheme, based on the distributed auction algorithm, that is robust to variations of the mmWave wireless channel and to mobility of client. In particular, the resulting optimal association solution does not have to be re-computed every time the network changes (e.g., due to mobility). Instead, the algorithm continuously adapt to the network variation and is thus very efficient. Numerical analysis verifies the ability of the proposed algorithms to optimize the association and to maintain optimality in dynamic environments of mmWave networks.