The Analysis And Research Of Distributed Coupling Simulation System Fault

With the rapid development of Internet technology, the new development trend of modern software system is larger and distribution. In the large distributed software systems, the data center or cloud computing center with thousands of computers, for example, due to the wide distributed resources, large number of nodes and complicated interaction, it is very difficult to develop, debug, optimize and maintain such kind of key support system, and the stability and reliability of the system faults’diagnosis and recovery becomes a huge challenge. Because the faluts and fails of the compute system’s hardware and software can’t be avoid, the result is that people would expect that their applications and a variety of computing system which they rely on is stable and reliable. Therefore, to response to this urgent need, the development and research of distributed system fault diagnosis technology becomes a focus in the various fields in recent years.The digital coupling simulation platform for dynamics of high-speed train is based on C/S mode of distributed simulation architecture, it primarily includes task scheduling, coupling controling, execute agency, and simulation of tasks, etc. The thesis takes the high-speed train coupling dynamics digital simulation platform as the research background, it studies and analyzes the potential failures of the system may encountered from job submission, job scheduling, job coupling calculation, etc. Analyze, detect and locale these failures, and take appropriate measures to deal with them. In this thesis, we put forward a series of algorithms such as job submission fault diagnosis strategy, job not scheduling fault diagnosis strategy, job suspend fault diagnosis strategy and job coupling calculation fault diagnosis strategy, etc. Job submission fault diagnosis strategy aims to location jobs submitted fault factors, automatic process those can be automatic processed failures. Thile the failures can’t be automatic processed would return the error code; Job not scheduling fault diagnosis strategy is to locate the reason that the job submission shows successful while the job scheduling does not conduct, and carry on corresponding fault processing. Job suspending fault diagnosis strategy is to locate the reason that simulation system suddenly stop and unable to continue simulating, and return an error code to tell the users; Communication failure in coupling calculation is the communication blocking and packet lossing occurred in simulation system coupling calculation between the coupler and the execution machine due to communication equipment failure or heavy Internet traffic. Coupling calculation communication fault diagnosis strategy achieves the simulation system breakpoint and continue calculating through mechanism such as information retransmission, timing detection, multiple requests, etc.Firstly, this thesis introduces the research background and significance, domestic and foreign research present situation and the primary research work. Secondly, the thesis presents the fault diagnosis and recovery strategy of job submission, job scheduling, job suspending and job coupling calculation communication. Finally, it summarizes and analyzes the entire work, and points out the thesis’s shortcomings and the further research work.