A system of systems is a set of heterogeneous independent systems that share data in pursuit of a common goal. These systems form a delay-/disruption-tolerant network (DTN), where routing is based on the store-carry-and-forward paradigm. Systems can communicate whenever they are close enough to each other, in what are called contacts. We assume that the movements of these systems may be predicted in advance and we consider that a sequence of contacts is given at the outset. During a contact, a given emitting system can transfer to a given receiving system a fixed amount of data (termed datum unit) that it has in its possession. The dissemination problem is to find a transfer plan such that all the data can be transferred from a given subset of source systems to a given subset of recipient systems. In this paper we study the problem where communications may fail. We propose an algorithm for finding a robust transfer plan that minimizes the dissemination length.