The vagaries of Indian monsoon rainfall have profound consequences, but improved forecasting, on all timescales, has proved elusive. Subseasonal variability significantly influences interannual variability. However, a wide range poorly understood processes and feedbacks control the variability. Systematic model biases persist, with moist convection a dominant source of error. Cumulus scale biases can grow upscale to affect synoptic scales in two days, and the largest scales in two weeks. The onset of the monsoon rains progresses northwards over India over about 6 weeks, the reasons for which are not well understood. This is shown, primarily using radiosonde data, to be related to a mid-level dry layer, which retreats in response to closely related thermodynamic and dynamic transitions. The dry mid-level layer does not, however, significantly control the interannual variability of the onset. Systematic rainfall biases in models with convective parametrisations typically include a Western Equatorial Indian Ocean (WEIO) wet bias, an Indian dry bias, and too weak lowlevel flow into India. Explicitly representing convection, in continental-scale simulations, is shown to reduce the dry bias and alters circulation and fluxes to sustain more rainfall over northern India. WEIO rainfall is too high in convection-permitting simulations with grid-spacings above ~2km, which also substantially alters the circulation. Towards a better understanding of the high rainfall upstream of the coastal mountains on the west coast of India, the role of offshore rainfall propagation is examined. Rainfall propagation likely results through storm advection and gravity waves, and varies in response to the passage of Boreal Summer Intraseasonal Oscillations. Contrary to satellite observations, rainfall consistently propagates onshore in convection-permitting simulations. The results demonstrate the importance of understanding scale interactions in the Indian monsoon. Improved models must capture storm-to-mesoscale processes that affect the entire monsoon.