Deriving physical parameters from gamma-ray burst afterglow observations remains a challenge, even now, 20 years after the discovery of afterglows. The main reason for the lack of progress is that the peak of the synchrotron emission is in the sub-mm range, thus requiring radio observations in conjunction with X-ray/optical/near-infrared data in order to measure the corresponding spectral slopes and consequently remove the ambiguity wrt. slow vs. fast cooling and the ordering of the characteristic frequencies. We observed GRB 151027B, the 1000th Swift-detected GRB, with GROND in the optical-NIR, ALMA in the sub-millimeter, ATCA in the radio band, and combine this with public Swift-XRT X-ray data. While some observations at crucial times only return upper limits or surprising features, the fireball model is narrowly constrained by our data set, and allows us to draw a consistent picture with a fully-determined parameter set. Surprisingly, we find rapid, large-amplitude flux density variations in the radio band which are extreme not only for GRBs, but generally for any radio source. We interpret these as scintillation effects, though the extreme nature requires either the scattering screen to be at much smaller distance than usually assumed, multiple screens, or a combination of the two.