The introduction of the next generations' mobile communications, 5G-Advance and 6G (5G-A/6G), promises boosting data throughput to new dimensions, achieving sub-millisecond latency, and providing wider coverage. Based on this promise, a great number of previously infeasible high-throughput, real-time, and IoT-based applications, such as high-resolution face recognition and extended reality, are being developed for deployment over 5G-A/6G networks. Such demanding applications assume that ample bandwidth will be available through the utilization of a high-frequency spectrum. However, at high frequencies, radio channels are susceptible to sudden changes in the surrounding conditions, generating highly fluctuating scenarios that directly impact the performance of upper-layer protocols and services. For applications that operate under end-to-end congestion control algorithm (CCA) (e.g., TCP- and QUIC-based applications), extreme fluctuations may generate unwanted behaviors that hurt the throughput and possibly favor non-CCA traffic with unfair results in bandwidth distribution. This paper thoroughly investigates the impact of fluctuating radio access channels on 5G-A/6G networks. We analyze the performance of various congestion control algorithms, including CUBIC, High-Speed, and BBR, as well as non-CCA traffic, under such conditions. Our evaluation, conducted through realistic simulations, examines the network's ability to maintain desired service levels amidst fluctuations. Furthermore, we explore the potential of state-of-the-art active queue management and buffer management policies at the gNB to mitigate the negative effects of these fluctuations and enhance overall network performance.