A distributed system task management system and method

Abstract

The invention discloses a task management system and method of a distributed system. The task management system comprises distribution nodes, a buffer pool, execution nodes, a one-sided operation module and a task tracking module, wherein the distribution nodes are used for distributing tasks to the buffer pool; the buffer pool is used for storing to-be-executed tasks; the execution nodes are used for acquiring tasks from the buffer pool for processing; the one-sided operation module is used for receiving user instructions and selecting an one-sided consumption mode or an one-sided production mode; the one-sided consumption mode refers to that stopping the distribution nodes from distributing the tasks to the buffer pool and allowing the execution nodes to acquire the tasks for processing; the one-sided production mode refers to allowing the distribution nodes to distribute the tasks to the buffer pool and stopping the execution nodes from acquiring the tasks from the buffer pool for processing; and the task tracking module is used for tracking the states of tasks and the information of the execution nodes where the tasks are. The task management system and method of the distributed system overcomes the defects that the conventional task manage process is insufficient in flexibility and robustness and seriously wastes resources.

Description

technical field [0001] The invention relates to the field of computer application technology, in particular to a task management system and method of a distributed system. Background technique [0002] Due to the limited capacity of a single system, the use of distributed systems is becoming more and more popular. For a distributed system, the entire life cycle of a task from dispatch to execution to summary of execution results may be distributed on different computing nodes in the system. And for a large-scale operating system, the system may contain tens of thousands of computing resources, and the number of computing nodes participating in task processing is extremely large. How to reasonably coordinate computing resources and how to track task execution is an urgent problem to be solved. However, the current distributed system still has the following problems: [0003] (1) The number and computing power of task dispatching nodes and execution nodes are not necessaril...

AI Technical Summary

Problems solved by technology

[0003] (1) The number and computing power of task dispatching nodes and execution nodes are not necessarily equal. There may be too many dispatching tasks leading to saturation. If you continue to dispatch tasks at this time, the system may be overloaded, resulting in downtime or system shutdown

[0004] (2) The distributed system involves a large number of computing resources. Due to function upgrades or operation and maintenance, some functional nodes may need to be shut down for maintenance, but the entire system is not expected to stop running, thereby reducing the cost of upgrades or operation and maintenance.

Images