Distance measurements are usually thought to probe the background metric of the universe, but in reality the presence of perturbations will lead to deviations from the result expected in an exactly homogeneous and isotropic universe. At least in principle the presence of perturbations could even explain the observed distance-redshift relation without the need for dark energy. In this paper we re-investigate a toy model where perturbations are plane symmetric, and for which exact solutions are known in the fluid limit. However, if perturbations are large, shell-crossing occurs and the fluid approximation breaks down. This prevents the study of the most interesting cases. Here we use a general-relativistic N-body simulation that does not suffer from this problem and which allows us to go beyond previous works. We show that even for very large plane-symmetric perturbations we are not able to mimic the observed distance-redshift relation. We also discuss how the synchronous comoving gauge breaks down when shell-crossing occurs, while metric perturbations in the longitudinal gauge remain small. For this reason the longitudinal (Newtonian) gauge appears superior for relativistic N-body simulations of large-scale structure formation.