Log detection method and device and log generation method and device applying computing power of intelligent computing center

Abstract

The invention provides a log detection method, a log generation method and a log generation device applying computing power of an intelligent computing center. The detection method comprises the following steps: receiving an anomaly detection instruction; in response to the exception detection instruction, obtaining an unstable log of a target software system specified by the exception detection instruction; extracting content features of each log message in the unstable log, and performing structured expression on the content features to obtain a log template; according to the time sequence of the log messages in the unstable logs, the log templates are aggregated to obtain a log sequence of the unstable logs; calling an unstable log anomaly detection large language model deployed in an intelligent computing center to detect the log sequence, and determining whether the running state of the target software system is abnormal or not; wherein the unstable log anomaly detection large language model is obtained by finely adjusting the basic log anomaly detection large language model by adopting historical unstable logs. According to the method, high-accuracy and high-timeliness anomaly judgment can be carried out on the unstable logs.

Images