Analyzing concurrent programs often involves reasoning about happens-before relations, handled by dedicated SMT theory solvers. Recently, preventative propagation rules have been introduced for consistency models to avoid unnecessary computations. This paper analyses the reproducibility of a recently published paper regarding a conflict-avoiding happens-before propagator. We show that the underlying axioms are insufficient for supporting sequential consistency. We find that the algorithm can leave out constraints on event ordering (even considering the original axioms), impacting the accuracy of verification. We show a simple counterexample to the stability claim in the paper. Two revisions of the algorithm are presented, and a proof on the correctness of these approaches respective of the original axioms is shown. The tool implementing the original algorithm is examined to ascertain how it circumvents wrong results. It is found that it deviates from the published algorithm. We show that an unmodified algorithm (via a patch in the implementing tool) causes incorrect results. We also show that our revised algorithm can be implemented efficiently in an independent verification tool.