Node management method and device

Abstract

The invention provides a node management method and device. The method comprises the following steps of obtaining a node in a response-stopping state; deleting a target file access path used for storing the node, and conducting resource clearing; if the node is in an active state, obtaining the target file access path again. According to the technical scheme, when the node is in the response-stopping state, instead of restarting the node, the target file access path used for storing the node is detected, resource clearing is conducted, and when the node is in the active state, the target fileaccess path is obtained again, so that the influences on other nodes are reduced, the high availability of a cluster file system is improved, the reliability of a virtualization platform is improved,and therefore the virtualization platform can have better stability in a bad network environment.

Description

technical field [0001] The present invention relates to the technical field of cloud computing, in particular to a node management method and device. Background technique [0002] Virtualization is one of the key technologies for building cloud computing infrastructure, and the cluster file system plays a key role in virtualization. The cluster file system provides massive storage for the virtualization platform and provides parallel access capabilities for virtual machines. [0003] The stability of the cluster file system directly affects the reliability of the virtualization platform. Compared with local file systems such as ext3, ext4, and xfs, the cluster file system provides functions such as lock protection and cluster heartbeat, which are more complex. However, in the actual application environment, many factors such as network state fluctuation among multiple nodes, node power failure, etc. will directly affect the HA (High Available, high availability) of the clust...

AI Technical Summary

Problems solved by technology

[0003] The stability of the cluster file system directly affects the reliability of the virtualization platform. Compared with local file systems such as ext3, ext4, and xfs, the cluster file system provides functions such as lock protection and cluster heartbeat, which are more complex. However, in the actual application environment, many factors such as network status fluctuations between multiple nodes, node power failures, etc. will directly affect the HA (High Available, high availability) of the cluster file system.

[0004] The second-generation Oracle cluster file system OCFS2 provides the cluster management function, which can be disconnected in the heartbeat, resulting in head-split (when the active node in the cluster stops responding for some reason, other nodes think it is dead and perform resource splitting). scramble, resulting in shared disk file corruption) or after the storage is disconnected, the faulty node is processed in time according to the arbitration policy, thereby reducing the impact on other nodes in the cluster; however, OCFS2 only provides two options for system panic reset and restart. This processing method has seriously affected the services such as virtual machines running on the virtualization platform, and greatly reduced the reliability and stability of the virtualization platform.

Images