Machine learning based instance remediation

摘要

Computer systems, such as network computing resources systems, are subject to hardware and software errors. To improve error handling and troubleshooting, information relating to errors is collected from a multitude of computer system and analyzed. As a result of this analysis, troubleshooting errors in computer systems is improved and errors are remediated automatically.

主权项

1. A computer-implemented method, comprising:
under the control of one or more computer systems configured with executable instructions,
for each computer system instance of a plurality of computer system instances:
detecting, based at least in part on console data output from the computer system instance, an error state for the computer system instance;for each user action of one or more user actions taken in connection with the computer system instance, recording the user action and a result of the user action in connection with the error state;using the recorded user action, result of the user action, and a machine learning algorithm to update a rules engine; andfor a particular computer system instance:
detecting, based at least in part on console data output from the particular computer system instance, a particular error state of the particular computing system instance;using the detected particular error state and the rules engine to determine one or more solutions for correcting the particular error state; andproviding the determined one or more solutions.