Monitoring Event Logs within a Cluster System

Resolving complex problems on cluster systems we have to take into account threats related to system dependability. We faced this problem in relevance to the developed project of the KASKADA platform targeted at managing heavy multimedia processing in a supercomputer environment (Galera cluster in Intel technology). Having analyzed the experience of other authors in the area of cluster dependability analysis and anomaly predictions we have found the need of developing appropriate strategy of monitoring the KASKADA platform within the Galera cluster. In the chapter we deal with the problem of exploring the event logs generated within the nodes of the platform. The main goal was to study the specificity of the platform operational image in the event space, explore the morphology and informative contents of logs and develop methods of anomaly detection. For this purpose special tools have been used.