Published February 4, 2026
| Version v3
Dataset
Open
Research Artifact – Cross-Project Flakiness: A Case Study of the OpenStack Ecosystem
Description
This repository serves as the research artifact for the paper “Cross-Project Flakiness: A Case Study of the OpenStack Ecosystem”. It contains a dataset of flaky tests mined from OpenStack, along with their associated metadata, scripts for data collection, and the analysis logic to reproduce the findings in RQs 1–3.
Requirements & Warnings
- Storage: This replication package is over 20 GB. Ensure you have sufficient disk space and a stable internet connection before downloading.
- Hardware: Due to the size of the datasets, high-memory (RAM) is recommended for running the analysis notebook.
- Database: A working installation of MongoDB and the mongorestore utility is required.
Replication Workflow
To replicate the results presented in the paper, follow these steps in order:
- Extract Logs: Unzip zuul_job_logs_2024_mark_filtered.zip into the project root.
- Restore Database: Use the mongorestore tool to import the study data from mongodb_archive.zip.
- Run Analysis: Execute scripts/BuildingResults.ipynb. This notebook contains all the logic to process the data and replicate the results, figures, and tables presented in the paper.
Note: scripts/detect_flaky_tests.py is the script used for daily data collection. You do not need to run this for replication, as the output is already provided in the output/ folders.
Repository Structure
scripts/ – Data Collection and Analysis
- BuildingResults.ipynb: The main Jupyter notebook used to replicate key results.
-
- Note: This script is resource-intensive and may take a significant amount of time to complete.
- detect_flaky_tests.py: The script used to collect flaky tests on a daily basis.
data/ – Dataset Files
- flaky_tests.json: All flaky tests identified in the study.
- cleaned_flaky_tests.json: Flaky tests that appeared in more than one patch set.
- flaky_builds_with_inline_comments.json: Flaky builds annotated with inline comments.
- flaky_jobs_with_inline_comments.json: Flaky jobs annotated with inline comments.
- inconsistent_flakiness_sample.xlsx: Manually labeled sample used for RQ3 analysis.
output/
- Output files generated by the detect_flaky_tests.py script.
materials/
- Visualizations and figures used in the paper.
survey/
- Questionnaire design and developer responses.
External Data Archives (stored separately)
- zuul_job_logs_2024_mark_filtered.zip: A collection of job logs gathered during the study.
- mongodb_archive.zip: A MongoDB database dump containing data used in this study.
Replication on the Zuul Ecosystem
- zuul_replication.zip: Scripts and data to replicate our results on the Zuul ecosystem.
Files
replication_package.zip
Files
(23.7 GB)
| Name | Size | Download all |
|---|---|---|
|
md5:4676e59b11e01f6287cbfa60c6ca3bcc
|
23.7 GB | Preview Download |