| With the rapid evolution of cloud computing deployment models at home and abroad towards a distributed form,the reliability of large-scale and complex networked software that is widely deployed in clouds and data centers is crucial.Among the many reasons that lead to such software failures and performance problems,user misconfiguration has become one of the most important causes,so pre-verification and positioning of its con-figuration defects is an important means to ensure the security of the current cloud and data centers.At this stage,the defect location technology for traditional software mainly faces two challenges when it is applied to the target software in this thesis.There are many configuration items,complex relationships,and the huge size of the module code,which makes it difficult to locate its defect configuration and code based on this module.In response to the above challenges,this thesis divides the entire location work into two parts:configuration defect verification and configuration defect location based on verification.Specifically,the core work of this thesis is as follows:(1)Through the analysis of the advantages and disadvantages of the existing config-uration error injection testing technology,it is found that the insufficient exploration of the configuration value space of the target software makes it difficult to trigger its con-figuration defects,which in turn leads to low efficiency of defect verification.Therefore,this thesis proposes an overlay mutation technology framework for target software con-figuration based on configuration item analysis,configuration item selection,and overlay mutation technology to improve the diversity and effectiveness of configuration value mu-tations,and to complete the verification through unit testing.Defective module location,while improving verification efficiency.Finally,with five softwares including Hcommon and HDFS as targets,relevant experiments are designed to verify the configuration item selection algorithm,superposition mutation algorithm and defect verification efficiency.Experimental results show that the superimposed mutation technology framework pro-posed in this thesis has better configuration defect verification efficiency,and at the same time achieves the positioning goal of the defect module.(2)Based on the results obtained in the defect verification stage,that is,the unit test of the mutated configuration item and the corresponding trigger target software defect,a configuration defect location technology based on incremental debugging is proposed,including a defect configuration location technology based on the binary reset algorithm And the two parts of the defect code location technology based on the execution compar-ison strategy,and then complete the defect code location through the defect configuration location results to improve the efficiency of configuration defect location.Taking the same five typical software as the configuration defect verification as the target software,relevant experiments are designed to verify the positioning efficiency of the binary reset algorithm and the execution comparison strategy.The experimental results show that the positioning accuracy of the two methods reaches 97.3% and 70%,respectively,and both have better performance in positioning speed.Through the configuration defect location technology in this thesis,7 of the 10 configuration defects of the target software can be located,and through this location information,this thesis also provides repair patches for them,and more than half of them have been officially confirmed. |
