Explainable GNN-based Approach to Fault Forecasting in Cloud Service Debugging

This repository contains supplementary materials for the paper:

R. Lovas, E. Rigó, D. Unyi, B. Gyires-Tóth

Explainable GNN-based Approach to Fault Forecasting in Cloud Service Debugging

If you need more data, code or information about our work, please contact the corresponding author: Róbert Lovas (robert.lovas (at) sztaki.hu)

How to Run the Experiments

First clone our repository:

git clone https://github.com/BME-SmartLab/GNN-Fault-Forecasting && cd GNN-Fault-Forecasting

Then build the docker image using our Dockerfile:

docker build -t gnn-fault-forecasting .

Next run the docker image interactively, mounting our repository into the container:

docker run -it -e USERNAME=myusername --gpus all --mount type=bind,source=$(pwd),target=/home/myusername/GNN-Fault-Forecasting gnn-fault-forecasting

And voilà, you can train the fault forecasting model or the fault explainability model:

python train_fault_forecasting_model.py
python train_fault_explainer_model.py

Citing

If you use our code or results, please cite our paper:

@article{
  R. Lovas, E. Rigó, D. Unyi, & B. Gyires-Tóth (2025). Explainable GNN-based Approach to Fault Forecasting in Cloud Service Debugging. IEEE Transactions on Network and Service Management.
}

Acknowledgement

The work reported in this paper, carried out as a collaboration between HUN-REN SZTAKI and BME, has been partly supported by the the European Union project RRF-2.3.1-21-2022-00004 within the framework of the Artificial Intelligence National Laboratory. This work was partially funded by the National Research, Development and Innovation Office (NKFIH) under OTKA Grant Agreement No. K 132838. The presented work of R. Lovas was also supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences. On behalf of the ‘MILAB - SmartLab' cloud project, we are grateful for the possibility to use HUN-REN Cloud which helped us achieve the results published in this paper.