An Automated Approach of Detection of Memory Leaks for Remote Server Controllers

Memory leaks are a major concern to the long running applications like servers which make the working set to grow with the program. This eventually leads to system crashing. This paper discusses a staged approach to detect leaks in firmware of remote server controller. Remote server controller monitors the server remotely with many processes running in the background. Any memory leak in the long running applications pose a threat to the performance of the system. The approach adopted here filters the processes running in the system with leaks based on time threshold in the first stage. These processes with leaks are passed to the next stage where precise memory leak detection is done using the open source dynamic instrumentation tool Valgrind. The system leverages an automated leak detection approach that invokes the leak detection process on encountering any severity in the system and generates a consolidated leak report. The proposed approach has less impact on the performance of the system and is faster compared to many available systems as there is no need to modify or re-compile the program. In addition, the automated approach offers an effective technique for detecting possible leakages in early software development phases.