| With the rapid development of software technology,the design requirement of software is more and more complexed.The use of traditional software development patterns for software development may lead to the following problems: from product development to publication,there may be a development environment,a test environment and a production environment are different;in the process of deploying software,it may cost too much resources;the period from product development to product delivery is too long.In the current Internet industry,more and more enterprises praise the idea of developing products rapidly.If we adopt the traditional way,it will be hard to satisfy.In order to solve the above problems,this paper designs and implements a continuous integration and publishing platform based on docker,which can help the development team to develop software efficiently and deliver product quickly.In this paper,the scheduling strategy of docker Swarm cluster is studied,and a dynamic scheduling strategy optimization algorithm is proposed to solve the problem that the same weight node may cause the node load imbalance.Through in-depth study of continuous integration and continuous publishing theory,the design objectives of the platform are put forward,the overall architecture of the platform is analyzed,and the platform is divided into modules,including core scheduling module,data persistence module,container management module and continuous integration module,and the continuous integration tools are also selected.Then,the platform is implemented,including the selection of the development environment and the implementation of the basic components.The components include the building of the docker Swarm cluster,the mirror image management,the container monitoring and the continuous integration.Finally,the platform is tested and analyzed,which proves the feasibility of the platform. |