The paper addresses the problem of scheduling tasks represented by a directed acyclic graph (DAG) on distributed environments. Due to the task, communication and resource constraints, the resource matching and task scheduling is NP-hard, even though the number of resources is abounded and task duplication is allowed. A new method named interpersonal relationships evolution algorithm (IREA) is given. The priority rules used are new, relationship number, potentiality, weight and merge degree are defined for cluster?s priority, and task potentiality for tasks? priority. The experimental results reveal IREA beats other five algorithms in terms of average performance, and it produces another optimal solution for the classic MJD benchmark.