Replayable Stack Trace for Investigating of Exception Occurrence Processes of A Java Web Application.

Abstract

When a runtime exception occurs during the operation or performance test of a Java Web application, it can not be confirmed detailed behavior and values until the occurrence of an exception from the limited log, therefore it takes time to isolate the problem. With the Selective Capture/Replay method, it is possible to reproduce and investigate the behavior and values of the modules at an arbitrary point in time, based on the execution history collected from the previously selected modules. However, due to the large load of externally outputting all the inputs of the modules, it need to select a small number of modules to be reproduced before the application runs, therefore it is not suitable for investigating unexpected runtime exceptions. In this paper, we propose “Replayable Stack Trace" as a low load capture method of execution history which can reproduce and investigate behavior and values up to exception occurrence for more modules. In the proposed method, the recent execution result of partial codes in a selected module is stored in a memory during executing the module,and when an exception occures, a stack trace with the stored values is externally outputted as execution history. By reproducing and showing the recent execution results of all the codes in each module based on the execution history, it is possible to confirm detailed behavior and values up to the occurrence of exception straddling the call hierarchy.