Distributed cluster state management method and system thereof

Abstract

The present invention is applicable to the field of computer technology, and provides a distributed cluster state management method and a system thereof. The method includes the following steps: A. setting an in-memory database and a manager of the distributed cluster; B. through the The in-memory database stores information of each node of at least one of the distributed clusters, and maintains information of each node of at least one of the distributed clusters through a manager of the distributed cluster. Thus, the management efficiency of the distributed cluster is improved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a distributed cluster state management method and system thereof. Background technique [0002] Currently, most cluster management systems are built on the basis of ZooKeeper. Such as Figure 1A Shown: Node A cluster, Node B cluster, and Node C cluster are all connected to the zookeeper cluster, and the entire distributed system is managed through Zookeeper, so that the entire system is in a usable state. [0003] Inside Zookeeper, Zookeeper maintains a virtual file structure for each node of the cluster. Nodes add nodes under their respective directories and watch the parent directory. After the node goes down, Zookeeper deletes the corresponding node from the directory. When there is a change in the current directory, Zookeeper notifies the nodes under the directory that there has been a change. as in Figure 1B When the node goes online, add data ( / APP3 / SERVER2) under it...

AI Technical Summary

Problems solved by technology

[0004] The defects of existing distributed cluster state management technologies include: 1. High degree of coupling and low degree of flexibility

Moreover, since the Zookeeper cluster needs a leader, when the leader node goes down, not only the data of the leader node is lost, but also the Zookeeper leader node needs to be re-elected to notify the node of the change. Data changes can only be synchronized from the Zookeeper leader node to the rest of the Zookeeper cluster nodes.

[0005] 2. The tree structure is cumbersome and complex, and the performance is low

[0006] 3. The rigid design of the "watch" mechanism: Zookeeper has no method support for obtaining the latest version information. It can only crudely register a watch every time it writes an updated method, and call back when these methods are called , it does not consider whether the content of the information has changed. Zookeeper will still call back for updates that do not change the content of the information, and the callback of Zookeeper is relatively rigid. It can only be called once. If the information continues to change, the watch must be registered again

[0007] In summary, the existing distributed cluster state management technology obviously has inconvenience and defects in actual use, so it is necessary to improve it

Images