Chain dependency analysis method and system for directed acyclic graph

Abstract

The present invention provides a directed acyclic graph chain dependency analysis method and system, wherein the method includes the following steps: using a global scanner to detect whether the previous batch tasks of the current batch have been completed; If the sub-tasks have been completed, instantiate all tasks in the current batch; run all root node tasks; judge whether all root node tasks are running normally; if all root node tasks have all run normally, pass the node scanner Monitor and run all subsequent tasks to which the respective root node tasks belong; if all the root node tasks of the current batch and the subsequent tasks to which they belong have all completed normal operation, the subsequent batches will be run. In the case of limited server resources, the present invention can efficiently analyze DAG chain dependencies and real-time task execution. When the current batch is running, only follow-up tasks can be monitored in real time through the node scanner, and its scanning operations are less, real-time Higher efficiency.

Description

technical field [0001] The invention relates to the technical field of block chains, in particular to a chain dependency analysis method and system for directed acyclic graphs. Background technique [0002] There are two solutions for the chain dependency analysis of directed acyclic graphs: one is to create a waiting task view, which monitors the task running table and task dependency table in real time, and adds tasks that can enter the running queue to the view Among them, the program scans the view at a certain time interval and runs the waiting tasks; the other is that the program scans the waiting tasks in the task running table in real time, analyzes whether the pre-dependencies of all waiting tasks have been completed, and runs the waiting tasks if completed . The problem in Solution 1 is: Since the analysis of view’s dependence on tasks takes a certain amount of time, the scanning time interval cannot be too small and difficult to predict, and the real-time perform...

AI Technical Summary

Problems solved by technology

The problem in Solution 1 is: Since the analysis of view’s dependence on tasks takes a certain amount of time, the scanning time interval cannot be too small and difficult to predict, and the real-time performance is relatively poor.

The problem in Solution 2 is: the number of tasks should not be too large, otherwise each polling of waiting tasks and analyzing the dependencies of waiting tasks will take a lot of time and reduce real-time performance; high-speed scanning will occupy a lot of server resources

However, regardless of scheme 1 or scheme 2, there is a problem that the scanning programs of the two schemes will always run regardless of whether the tasks have been completed, which will cause a waste of server resources.

Images