Method and device for split-brain prevention based on disk service lock

Abstract

The invention discloses a method and a device for preventing split brain on the basis of a disk service lock, belongs to the technical field of computer cluster and is suitable for a high availability cluster. The method and the device for preventing the split brain on the basis of the disk service lock avoids the problem that when a cluster heartbeat network is interrupted, states of other nodes and operation service of the nodes cannot be judged accurately, service cannot be taken over, and the split brain is hard to avoid especially under the condition that the technology for restarting or storing and isolating a system is not used. The technical scheme includes: when the heartbeat network is interrupted, whether the heartbeat network is broken down or the nodes are halted is judged effectively; whether smooth stopping of service can be obtained correctly through inquiry of the disk service lock so as to avoid the problem of short split brain; a plurality of nodes grab the service lock simultaneously, only the successful node can start service to avoid split brain.

Description

[0001] (1) Technical field [0002] The invention belongs to the technical field of computer clusters and is applicable to high-availability clusters, in particular to a split-brain prevention method and device for a high-availability system based on a disk service lock. [0003] (2) Background technology [0004] With the rapid development of communication network technology, key fields such as telecommunications, finance, and e-government have higher and higher requirements for server availability, and high-availability cluster systems are increasingly accepted and widely used by more and more users. In order to avoid the huge economic loss caused by the failure of the server or switch to stop the business, the high availability (High Availability, HA) system can greatly reduce the routine maintenance of the business system and the downtime of the business due to software and hardware failures. [0005] The current high-availability cluster system is mainly connected through ...

AI Technical Summary

Problems solved by technology

However, the inventors found that these methods cannot meet the requirements of the market. In actual implementation, other important services are also running on the standby node, and the customer does not allow the operating system to be restarted or the entire shared storage to be isolated

[0008] In the process of solving the cluster to prevent split-brain, the inventor found that the existing technology has the following problems: when the cluster heartbeat network is interrupted, it is impossible to distinguish whether the heartbeat network is faulty or the node is dead, and it is easy to make wrong judgments, resulting in split-brain ;In the above judgment, it can be detected that the working machine is still alive, but it cannot be determined whether the service has stopped successfully, and there is a possibility of a short-term brain split, resulting in data corruption; in the above judgment, if it is judged that the node is down, there are multiple nodes that start at the same time service, resulting in a split brain, resulting in the possibility of data being destroyed

Images