Design And Implementation Of Hardware Monitoring System In Cluster Environment

The scientific information product processing subsystem is a part of the earth's surface system science research application demonstrate project which is developed by college of computer and information engineering in Henan University entrusted by remote sensing and digital institute of Chinese Academy of Sciences.The project,which contains a large amount of computer hardware,works in the cluster system.The hardware monitoring system in the cluster environment,which monitors hardware information and the connection status of each node and then these information are showed to users,is designed for the system so as to lower the difficulty of cluster management,reduce the problems caused by hardware failure and help the manager manage the cluster expediently.The hardware monitoring system in cluster environment is divided into three functional modules including information acquisition,information collection and information display.(1)The hardware information and running condition of each node is obtained by the information acquisition module and then sent to the server.(2)The information collection module chiefly parses XML information,saves them in the server and then sends to the client.(3)The information display module mainly shows hardware information which is monitored to users in the client.The Sigar class library is obtained to monitor hardware information and running condition of each node,DOM4 J is used for parsing the transferred XML information.CPU and memory usage records are showed in the form of dynamic line graph using JFreeChart technology,flexible remote calls as well as the distributed object technique for callback of RMI enable communication between modules.This paper designs and achieves the hardware monitoring system under cluster environment,which has achieved expected effort in the process of experimental operation with low occupancy resources and sensitive reaction,is able to effectively monitor hardware information and running connection condition of each node in the system.The users may observe and analyze the entire cluster,discover problems in time which reduce the loss in cluster system due to hardware failure.The efficiency of cluster manager is improved in the scientific information product processing subsystem project.