A cluster task management system

Abstract

The invention discloses a cluster task management system, comprising: a task scheduling center receives an operation request from a user, generates a corresponding task instruction according to the operation request, sends the task instruction to a distributed cache middleware, and receives the distributed cache middleware to return task processing information; the distributed cache middleware receives task instructions, sends the task instructions to the task processing module, receives the task processing information returned by the task processing module, and sends the task processing information to the task scheduling center; the tasks set on each node The processing module receives task instructions, performs corresponding task processing according to the task instructions, generates task processing information, and sends the task processing information to the distributed cache middleware. The system transforms the task of cluster management from centralized to decentralized. Through the cooperation of task scheduling center, distributed cache middleware and task processing modules installed on each node, it realizes the decentralization of cluster management and balances the cluster load.

Description

technical field [0001] The invention relates to the field of cluster technology, in particular to a cluster task management system. Background technique [0002] A cluster is a group of independent computers interconnected by a high-speed network that form a group and are managed as a single system. Through cluster technology, relatively high benefits in performance, reliability, and flexibility can be obtained at a low cost, and cluster applications are very extensive. [0003] Task scheduling in the cluster is the core technology. The main purpose of the cluster is to provide services as a whole, and the extra burden caused by the management of the cluster cannot affect the normal business of the cluster. [0004] In existing cluster-level management, a single master node is usually selected, and the master node issues and processes tasks. In this mode, in the case of business load balancing, the master node has one more load than other nodes, that is, the management lo...

AI Technical Summary

Problems solved by technology

In this mode, in the case of business load balancing, the master node has one more load than other nodes, that is, the management load. When the management load is too high, it will cause the master node to be overloaded compared to other nodes. In the case of unbalanced cluster load, the unbalanced load will easily cause the master node to process slowly and easily crash. Once the master node crashes, the management of the entire cluster system will be paralyzed

Images