Preserving inter-actor connectivity is essential in most wireless sensor and actor network (WSAN) applications as nodes have to collaborate and coordinate their actions against the events reported by the sensors. However, failure of a critical (i.e., cut-vertex) node partitions inter-actor network into disjoint segments and thus hinder network operation. The prime objective of this paper is to analyze the performance of reactive connectivity restoration algorithms for delay-tolerant WSAN applications. First, we provide insights to the state-of-the-art reactive connectivity restoration algorithms. Then, we present a Nearest Non-critical Neighbor (NNN) algorithm; a localized and distributed reactive approach for reconnecting network partitions. In NNN, each actor periodically determine its criticality (i.e., cut-vertex or not) based on 2-hop information and exchange with its neighbors. In case of a critical actor failure, the neighbors detect and trigger a connectivity restoration procedure that involves controlled and coordinated node relocation. NNN prefer to displace non-critical nodes during relocation in order to minimize recovery overhead in terms of distance movement and message coordination. We analyze the performance of reactive schemes through theoretical analysis and simulations.