Fault-tolerant mechanism of both job execution and file transfer for integrated nuclear energy simulation

Tatekawa, Takayuki; Teshima, Naoya; Suzuki, Yoshio   ; Takemiya, Hiroshi

In the nuclear field, various large-scale integrated simulations which cannot be executed with single job have been developed to reveal physical and engineering phenomena. Such integrated simulations are accomplished by coupling several simulation codes, each of which is charge of each physical process or each engineering part of whole system. Fault-tolerant (FT) mechanism is very important to run such simulations on the error-prone environment such as Grid. We developed functions of error detection, job re-submission, and file re-transfer and integrated them as a FT mechanism. Our test run of integrated nuclear energy application showed that the FT mechanism sustained the long run of the application by recovering the job failure automatically.