Monitoring Anomalous States of Computer Systems by Intellectual Analysis of Data of System Journals
Oleg I. Sheluhin, D. Kostin, Mikhail Gorodnichev
Abstract
The ways and algorithms for automating the control of computer systems by intelligently analyzing unstructured data from system logs in order to detect and diagnose abnormal states are analyzed. This information is needed by the technical support service to localize the problem and accurately diagnose it. Because of the ever-increasing size of logs, intelligent analysis data models are being used to help developers extract system information. The research was conducted on a network infrastructure consisting of 6 hosts that make up a cluster under Rancher management. Expertly, events that significantly affect the system and are abnormal, related to problems with the passage of traffic on the network, or denial of service. Of the 988 message templates, 81 patterns are marked as abnormal.