Task scheduling method, device and control node

Abstract

The invention provides a task scheduling method and device as well as a control node. The task scheduling method comprises the following steps: responding to triggering of a task, acquiring first calculation parameters of each computing node connected to the control node at fixed acquisition time intervals; after acquiring the first calculation parameters for first preset acquisition times, calculating a next acquisition time point with the reference of the initial CPU usage rate of each computing node as well as an initial CPU usage rate sequence and an initial memory usage rate sequence which are respectively composed by the initial CPU usage rates of the computing nodes; according to each of schedule variation sets respectively composed by the calculating tasks, judging whether each schedule variation set represents that the corresponding computing task is executed slowly or not; if yes, eliminating the computing task which is executed slowly, and reallocating another computing node; if not, acquiring first calculation parameters of each computing node connected to the control node again. According to the task scheduling method and device as well as the control node, the task executing efficiency can be improved, and computing resources are saved.

Description

technical field [0001] The present application relates to the technical field of computer data processing, in particular to a task scheduling method, device and control node. Background technique [0002] Distributed computing clusters can be divided into distributed real-time computing clusters and distributed offline computing clusters. In the existing distributed offline computing clusters, there are two basic roles, namely control nodes and computing nodes. The control node can assign tasks to each computing node, and the computing nodes execute each task. In practical applications, computing nodes may experience unexpected situations due to abnormal reasons. In order to ensure that the cluster can still provide external services after a single node fails, the control node can assign tasks on the failed computing node to other computing nodes. restart on the node. [0003] However, the inventor discovered during the research process that the computing task has been run...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by this application is to provide a task scheduling method to avoid the situation in the prior art that a task still needs to be re-run in a reassigned computing node after it has been executed for a period of time, so as to solve the problem of repeated computing tasks. The problem of wasting computing resources during execution

Images