The Design And Implementation Of High-availability Cluster Management And Monitoring System Based On Linux

With the development of technology, Computer is playing a more and moreimportant role in People’s Daily life. At the same time, with the development oflarge computer cluster, the cluster node failure about computer has become anunavoidable problem.This article implements a high availability cluster system which manage andmonitoring the cluster. And this system provide a simple way to control the cluster,system administrator visits the web console and acquire the information of clusterperformance monitoring data. Through the technologies such as heartbeat detection,fault detection and Private IP, this system implements the function which standbymanagement node can take over the master’s authority by the fault event. Theweb-end of this system use the framework of Django which based on MTVarchitecture and use JavaScript, HTML and CSS to accomplish the exhibition webpage at the same time. The Implements of back-end use Python to unify thelanguage of fore-end and back-end.The core business logic layer of this system is divided into five modules, nodemanagement module, database management module, information collection anddelivery module, middleware module and log management module. Nodemanagement module is responsible for the node management function which controlall the nodes and network, and meanwhile it also provide identity and accessmanagement, fault detection and processing, heartbeat detection and processing.Database management module is responsible for managing the cluster databaseoperations and providing fault detection and processing for the database.Information collection and delivery module is responsible for providing thecollection and persistence of performance information for the cluster and it can sendthe alarm mail according to the unusual of performance information. Middlewaremanagement module provides the management function of RabbitMQ messagecluster. Log management module provides the function such as the storage, theunloading, the cleaning of logs for cluster.With the functional and non-functional testing, we confirm that the system canmanage and monitoring the cluster. And system has expected the functional andhigh availability requirement. The whole system has been put into practical used.