| The high availability of cluster management system refers to providing continuous services to the external. This thesis aims at the high availability of cluster management system, treats the open source cluster provision and management software KUSU as the basis, and sees the dual hot backup theory of the cluster system’s management node as the support. Also, the thesis regards the cluster system fault detection and automation of business switch as the core. Taken the above into account, the thesis researches and realizes the high availability cluster management system based on Linux platform.The system, based on KUSU, has been redeveloped with adding mechanisms of heartbeat detection, service switching, data synchronization, floating IP address and the switching between Installer Node and Failover node, to achieve the high availability cluster management system with fault detecting and handling, and this is the main research content of this thesis. Heartbeat is used for communication between managing nodes and their backups in order for monitoring their healthy status. At the same time, the system uses PostgreSQL, a database management software, which has the ability for reducing crush, to achieve data synchronization. Moreover, Pacemaker has also been used, and this is the technique to take over resources and service switching, thereby there would be no break of service when the switch between managing nodes and backups is necessary. In addition, in order to provide smooth access of services, the system uses ARP cheat to achieve the floating IP address shared between Installer Node and Failover node. Furthermore, according to the needs of the practical application scenarios, the system provides the two switching way which is automatic switching and manual switching.This thesis will elaborate the system’s key issues and its implementation, as well as the overall system design and implementation process. Finally, test cases and test results in the physical environment are given. |
PDF Full Text Request