Isolation method and device in a cluster system

Abstract

The present invention provides an isolation method and device in a cluster system. The method includes: receiving a fence operation command sent by the main management node of the cluster; the fence operation command carries target physical host node information and target application program information; if possible Execute the fence operation command received from the cluster master management node locally, then determine the fence device corresponding to the fence operation command, trigger the fence device through the interface corresponding to the fence device, and execute the target physical host through the fence device The fence operation of the target application on the node. The invention can guarantee the safety and consistency of shared data.

Description

technical field [0001] The invention relates to cluster technology, in particular to a fence method and device in a cluster system. Background technique [0002] High Availability (High Availability) cluster is a group of computer systems that jointly provide network resources for clients, providing transparent redundant processing capabilities, so as to achieve the goal of uninterrupted applications. Each of these computers providing services is called a physical host node (Node). When a physical host node is unavailable or unable to process the client's request, the request will be transferred to another available physical host node for processing in time, and these are transparent to the client, and the client does not need to care about the specific location of the resource to be used, the cluster The system will do it automatically. [0003] In a high-availability cluster environment (such as a cluster system based on Heartbeat or Corosync), it is necessary to protect...

AI Technical Summary

Problems solved by technology

[0003] In a high-availability cluster environment (such as a cluster system based on Heartbeat or Corosync), it is necessary to protect the security and consistency of shared data, otherwise, it will cause serious problems and even cause system crashes

For example, if an application (such as a virtual machine) on a physical host node X in the cluster is abnormal and is in an unavailable state, the cluster management will detect the abnormality and start the application on another physical host node Y , if the application on the physical host node X is not isolated in time at this time, the two applications will access the same resource, such as a shared database or a shared disk, and if a write operation occurs, data inconsistency will result

Images