This paper proposes and analyzes the performance of two simple transmission protocols for a multiuser massive multiple-input multiple-output relaying system, where K single-antenna users transmit data to a massive-antenna destination through an $N$ -antenna relay node. The proposed transmission protocols take imperfect channel state information (CSI) and power allocation into consideration. In the first transmission protocol, the relay does not need to know the CSI. It just amplifies and forwards the received signals to the destination ultilizing the knowledge of the large-scale fading coefficients. In the second protocol, the relay first estimates the channels from all users in the training phase. With these channel estimates, it then uses the classical maximum-ratio combining (MRC) technique to combine all received signals and forwards them to the destination. In both protocols, the destination estimates the channels and employs MRC to decode the signals transmitted from all users. We propose an efficient channel estimation method at the destination in which the destination estimates only the effective channels gains, instead of the whole channels. As a consequence, the channel estimation overhead does not depend on the numbers of relay and destination antennas which makes the system scalable. We derive closed-form expressions for the spectral efficiency of the two transmission protocols. These results allow us to further analyze the performance of these two schemes and to meticulously allocate the transmit powers. Particularly, a max-min power control algorithm is proposed which selects the transmit powers at relay and the users to maximize the lowest spectral efficiency of all users for each large-scale fading realization. We show that, by using the max-min power allocation algorithm in both schemes, the spectral efficiency can be increased significantly, compared to uniform power allocation. Furthermore, a comparison between the two transmission schemes is pursued. If the distance between users and the relay is large, the first transmission protocol is better and vice versa if this distance is small.