This study experimentally demonstrates a novel blockage prediction based on exhaustive beam-pair scan in millimeter wave (mmWave) networks toward 6G. MmWave wireless systems face the challenge of susceptibility to link interceptions. When the line-of-sight (LOS) path is blocked by obstacles such as pedestrians, the received power drops sharply. The prediction uses the signal strength of all transmit/receive beam-pairs through a two-sided scan rather than that of all transmit beams through the one-sided scan assumed in existing studies. The time series of all transmit/receive beam-pairs is informative with respect to the location and velocity of the obstacle. Thus, it might be possible to predict future LOS blockages. The mapping from the signal strength of all beam pairs to future blockages was predicted with machine learning. The experimental results validate that the blockage prediction through a two-sided scan of all beam pairs is more accurate than that through a one-sided scan.