A large-scale concurrent evaluation method for programs based on container technology

Abstract

The invention relates to a program large-scale concurrency evaluation method based on container technology. The metho aims at the shortcomings of the existing procedure evaluation technology, decomposes the whole evaluation process reasonably and effectively, performs modular and flowing design, and uses containers to provide a running environment for program evaluation, uses Kubernets for container cluster management, builds a private mirror warehouse, evaluates container required and Gitlab server, constructs a set of evaluation script templates. by generating the corresponding evaluation script, evaluating the execution container environment pod, pulling the evaluator to execute the evaluation script, finalize the evaluation of the program and return the results. The method effectivelysolves the problems of security, resource quota and allocation, user space isolation and automation of traditional evaluation methods, and realizes the automation of evaluation process through reasonable process design, and realizes the stable and efficient support of large-scale program evaluation through the personalized setting of container clusters.

Description

technical field [0001] The invention belongs to the technical field of program evaluation application, and in particular relates to a large-scale concurrent program evaluation method based on container technology. Background technique [0002] Against the background of the rapid development of the Internet, the importance of computer programs is becoming more and more significant. Program evaluation, as an important means of testing program correctness, efficiency and other attributes, is used more and more widely and deeply in the fields of software development, code analysis and online programming. Code evaluation is directly related to product quality, life cycle and cost in software development. In the field of online programming, online coding platforms such as LeetCode, Lab Building, and CodeFights directly rely on code evaluation to analyze the user's code and give feedback. Similarly, in the field of code analysis, various academic research fields such as fuzz test...

AI Technical Summary

Problems solved by technology

[0003] Traditional program evaluation needs to complete operations such as source code editing, program compilation, execution, input and output, and indicator analysis on physical hosts, virtual machines, or cloud hosts. Such a complex process has obvious defects

[0004] The first is security issues, mainly including resource security and system security. Users have different abilities, and the specific behavior of the code they write is uncontrollable during execution. Dangerous actions such as CPU resources, deletion of host machine files and directories, malicious deletion or stopping of host host key services, malicious Trojan horses, and virus implantation are likely to seriously affect the normal operation of the host machine itself;

[0005] The second is the performance problem. The resource consumption of a single evaluation request is not enough to affect the host machine. However, in some cases, such as online programming platforms, there may be a large number of users online at the same time, and a large number of users may evaluate at the same time. In this case Under the circumstances, a large number of evaluation requests will put a huge pressure on the memory and CPU resources of the host machine and even cause the host machine to crash. In the case of a single machine, it is more complicated and costly to expand the capacity of the evaluation system;

[0006] Another issue is concurrency and resource isolation. When a large number of evaluation operations are performed concurrently, the construction system and some resources of the system will be shared. Therefore, problems such as resource competition and deadlock will occur on the host machine. At the same time, due to the inability to perform resource isolation, Different evaluation operations may also affect each other and affect the final evaluation results;

[0007] Finally, there is still the problem of automation. The evaluation process is a process that includes several steps and must be strictly coherent. Whether the evaluation can be automated and streamlined is directly related to the evaluation efficiency.

Images