This article concerns the decentralized adaptive tracking controller design problem for a class of output-constrained uncertain nonlinear interconnected large-scale systems under Markov jump actuator faults. In view of the intermittence, time-variability and randomness of faults, a group of time-varying Markov variables is introduced to characterize the actuator faults in each subsystem. First, a hyperbolic tangent type auxiliary function is constructed to compensate subsystem interactions of unknown strengths by using local information only. Second, adaptive mechanisms are developed to estimate the bounds of unknown faults, and a decentralized fault-tolerant controller is proposed. Meanwhile, the output transformation strategy is used to constrain the subsystem output, thereby enlarging the range of the initial output value in comparison with the barrier Lyapunov function based control method. With the proposed scheme, all the closed-loop signals are bounded in probability and the output constraint is not violated in each subsystem. Finally, simulation studies are conducted on a double-inverted-pendulums system connected by a spring to demonstrate the effectiveness of the proposed control scheme.