Lazy scheduling, i.e. setting transmit power and rate in response to data traffic as low as possible while satisfying delay constraints, is a known formulation of energy efficient transmission. Solutions exist for offline and infinite-horizon online versions of the problem. This paper addresses the finite horizon online transmission scheduling problem under stochastic packet arrival, energy harvesting and channel variation processes. The main contribution is a mechanism to obtain an online algorithm using explicit closed form expressions of the derived offline optimal policy. The resulting low complexity online algorithm attains near-optimal performance, not only asymptotically, but also in the practically interesting short term.