AFETM: Adaptive function execution trace monitoring for fault diagnosis

TL;DR

This paper presents AFETM, an adaptive function-level fault diagnosis method that reduces overhead and improves accuracy by combining dynamic tracking, fault injection, and graph neural networks.

Contribution

It introduces a novel adaptive function tracking approach with techniques for trace point selection and graph construction, enhancing fault diagnosis effectiveness.

Findings

Outperforms log-based and full tracking methods in accuracy

Reduces overhead compared to traditional methods

Improves fault diagnosis performance on web service benchmarks

Abstract

The high tracking overhead, the amount of up-front effort required to selecting the trace points, and the lack of effective data analysis model are the significant barriers to the adoption of intra-component tracking for fault diagnosis today. This paper introduces a novel method for fault diagnosis by combining adaptive function level dynamic tracking, target fault injection, and graph convolutional network. In order to implement this method, we introduce techniques for (i) selecting function level trace points, (ii) constructing approximate function call tree of program when using adaptive tracking, and (iii) constructing graph convolutional network with fault injection campaign. We evaluate our method using a web service benchmark composed of Redis, Nginx, Httpd, and SQlite. The experimental results show that this method outperforms log based method, full tracking method, and Gaussian influence method in the accuracy of fault diagnosis, overhead, and performance impact on the diagnosis target.