Dockerfile TF Smell Detection Based on Dynamic and Static Analysis Methods
Jiwei Xu, Yuewen Wu, Zhigang Lü, Tao Wang · 2019
Dockerfile is used to build docker image. In the image building process, temporary files are frequently used to import applications and data. A careless use of Dockerfile may cause temporary file left in the image, which can increase the image size, thus effects the elastic scale ability and QoS. This problem is identified as temporary file smell. The feature of UnionFS that docker image used is different from traditional filesystems. If users are not paying attention, they are too apt to make such mistakes. To address this problem, we propose two different methods to detect temporary file smell with dynamic analysis and static analysis respectively. We use the really-world cases to evaluate our methods. Experimental results show that our methods can effectively detect the temporary file smell.