Finding Errors in Massive Payment Log Data
Year: 2017
Type: dissertation
Abstract: Logs play an important role in debugging and maintaining large applications. When a system fails, developers investigate the log records to gain insight to identify the problem. Traditionally developers used tools like grep and tail to identify irregular behavior. This approach is time consuming for large companies that can generate over 600GB of log data each day. We propose Logness, an approach that clusters raw log data and uses developer feedback to prioritize important log messages. Through experiments on live log data at Adyen, we show that our approach is able to capture subtle, but serious mistakes from the log data. In this thesis, we also share some of our success stories and lessons learned while applying Logness in the wild.
Author: P. Evers
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2604960690