Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/116387
|Automated characterization of build and test failures on a continuous integration system
|Esquembri Moreno, Gerson
Caballero González, Carlos
|Continuous integration systems allow for easy identification of build and test failures in software projects. However, in projects with many developers and deliverables, where the same code is used with different configurations for many different products, it is not easy to identify what changes in the code caused the failures. This usually results on involving a team to make sure that the failures are identified as soon as possible so the broken code can get reverted, and it does not affect future rounds. Early failure detection prevents the products which are going to be released to be faulty, and helps developers work with a stable codebase that will help them identifying the result of their modifications without the noise generated by other developers' issues. A way to solve this problem is developing a strategy that allows the build and test failures, and their root causes, to be identified automatically. The strategy must be based on the identification of possible candidates for the broken builds/tests, evaluating the changes between the last successful integration and the failing one. Furthermore, a strategy to evaluate if the offending commit (the one which caused the failure) should be automatically reverted or not could be developed, as well. The state of this project after the development phase demonstrates the possibility of developing a system capable of automating failure detection on a complex integration system, where manually finding the responsible changes of the failures can take from a few minutes from a few hours, meaning that the figure of a full time system integrator is needed. With the automated failure detection tool, finding the offending commits takes just a few seconds, making it much easier for the system integrator to revert the changes that caused the failure.
fallos de desarrollo
|Appears in Collections:
|Trabajos finales de carrera, trabajos de investigación, etc.
Files in This Item: