With the popularity of cloud computing, its reliability is becoming a key concern for cloud users. As the main product of IaaS cloud, virtual machine's reliability is a valuable research issue. This paper presents a modeling strategy for VM reliability based on general stochastic Petri nets. The modeling strategy decomposes VM reliability into network reliability, computing reliability and storage reliability, and considers high availability mechanism in IaaS cloud. To analyze the reliability model, we use Markov process theory and give the detailed solving process. A case study of OpenStack cloud is conducted to illustrate the applicability of the proposed modeling strategy. In addition, we discussed the reliability impact of some model parameters which can help IaaS cloud on how to improve VM reliability.