Recent observations of Type II supernovae have revealed that their red-supergiant progenitors lose a significant amount of mass during the last years of their evolution. However, because it is difficult to discover supernovae within days of explosion, the diversity of mass loss in red supergiants has not yet been fully mapped. This talk presented the case of SN 2021yja, which was serendipitously imaged within hours of explosion and observed with a sub-day cadence during its rise to peak. From the exceptionally long plateau period and the high nickel mass, we infer a relatively massive red-supergiant progenitor star. However, archival imaging from the Hubble Space Telescope places a stringent upper limit of on its progenitor mass. We discuss these conflicting constraints in the context of the larger sample of exploding red supergiants. Our analysis helps illuminate the poorly understood mechanism(s) behind red-supergiant mass loss.