Datasetpublic

toolathon123/workflow-failure-triage

Workflow Failure Issue Triage This dataset contains CI/CD workflow failure issues for the past month. This README serves as the auto-generated summary report. Executive Summary Reporting period: Mar 03, 2025 – Apr 01, 2025 — 30 issues logged; 25 valid issues analyzed (Mar 03, 2025 – Mar 31, 2025). Excluded: 5 issues (3 closed, 2 duplicates) removed from analysis. Top root cause: Test Timeout (6 issues, 24% of valid), followed by Build Failure and Deployment… See the full description on the dataset page: https://huggingface.co/datasets/toolathon123/workflow-failure-triage.

sourceHugging Faceupdated 2mo agoView on Hugging Face

0likes13downloads

Dataset Card

Workflow Failure Issue Triage

This dataset contains CI/CD workflow failure issues for the past month. This README serves as the auto-generated summary report.

Executive Summary

—Reporting period: Mar 03, 2025 – Apr 01, 2025 — 30 issues logged; 25 valid issues analyzed (Mar 03, 2025 – Mar 31, 2025).
—Excluded: 5 issues (3 closed, 2 duplicates) removed from analysis.
—Top root cause: Test Timeout (6 issues, 24% of valid), followed by Build Failure and Deployment Error (5 each).
—Volume: failure flow was steady at ~6 issues/week across the four full weeks, then dropped to 1 in the final partial week.
—Attention: 5 issues (20%) could not be classified (e.g. "Random pipeline crash", "Unexpected error") — consider extending the triage rules to capture these.

Classification Counts (25 valid issues)

Root Cause	Count	Share of Valid
Build Failure	5	20%
Test Timeout	6	24%
Deployment Error	5	20%
Configuration Error	4	16%
Unknown	5	20%
Total	25	100%

Weekly Breakdown

Week (Mon–Sun)	Build Failure	Test Timeout	Deployment Error	Configuration Error	Unknown	Total
Mar 03 – Mar 09	1	3	1	0	1	6
Mar 10 – Mar 16	1	1	1	2	1	6
Mar 17 – Mar 23	2	1	1	1	1	6
Mar 24 – Mar 30	1	1	2	1	1	6
Mar 31 – Apr 06 (partial)	0	0	0	0	1	1
Total	5	6	5	4	5	25

Week-over-Week Observations

—Test Timeout spiked in the week of Mar 3 (3 issues), then settled to a steady 1/week — the most recurring failure category.
—Build Failure peaked in the week of Mar 17 (2 issues); otherwise 1/week.
—Deployment Error peaked in the week of Mar 24 (2 issues); otherwise 1/week.
—Configuration Error peaked in the week of Mar 10 (2 issues), then held at 1/week.
—Unclassified issues remained a constant ~1/week throughout — a persistent triage gap worth addressing.

Excluded from Analysis

Issue	Title	Created	Status	Reason
WF-005	Build failure on main	2025-03-07	duplicate	Duplicate of another issue
WF-012	Deploy failed due to config	2025-03-14	closed	Already closed / resolved
WF-018	config_error: missing env	2025-03-20	closed	Already closed / resolved
WF-025	build-failure on staging	2025-03-27	duplicate	Duplicate of another issue
WF-030	Build failed for feature X	2025-04-01	closed	Already closed / resolved

Excluded total: 5 issues — 3 closed, 2 duplicates. Exclusions are applied to all counts above.

Methodology

—Failure categories were normalized with fuzzy matching per rules.txt (e.g. build_failure, build-failure, build failure → Build Failure).
—When a category was missing, the title was classified by keywords (build, timeout, deploy, config); otherwise the issue was marked Unknown.
—Issues with status closed or duplicate were excluded from all counts.