具有容错机制的机群监控系统  

Fault-tolerant cluster monitoring system

作　　者：刘青昆[1] 孙爱婷[1] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081

出　　处：《计算机工程与设计》2010年第21期4599-4602,4624,共5页Computer Engineering and Design

基　　金：辽宁省博士科研启动基金项目(20051058)

摘　　要：通过对大规模机群监控系统可靠性分析,提出了一种容错的机群监控节点模型。该模型针对Ganglia机群监控系统在处理监控节点故障时存在的问题,提出了将Ganglia机群监控技术和心跳技术相结合的解决方法,通过修改与两种技术相关的配置文件,实现了二者的无缝连接,有效地处理了监控节点故障的转移和恢复。实验结果表明,该模型避免了机群监控节点故障,增强了监控系统的可用性,提高了无人值守的可信度。After analyzing the reliability of large cluster monitor system, a kind of fault-tolerant cluster monitormg none mooel is put forward. To solve problems of Ganglia monitoring node fault, a new method is proposed based on the Ganglia and Heartbeat. By modifying their configuration files, seamless connection is implemented, and the failover and recovery of cluster monitoring node fault is dealt with. This model avoids the failure of the cluster monitoring node, enhances the availability of cluster monitoring system and advances the reliability of nobody hold monitoring.

关 键 词：机群监控 可靠性 容错 监控节点故障 心跳技术 

分 类 号：TP302.8[自动化与计算机技术—计算机系统结构]