CI Failure Triage
Design a build-results view that helps a developer distinguish code regressions, flaky tests, and infrastructure failures.
The brief
Understand the problem
Background
A failed pipeline may contain thousands of log lines and several downstream cancellations. Repeated reruns can waste capacity and hide a real regression when the likely cause is unclear.
User context
Noor opens a pull request whose integration build failed after a database test timed out. The same test passed on the main branch and failed twice this week on unrelated changes.
Product problem
Developers need a concise failure diagnosis, supporting evidence, ownership, and safe next actions without treating a likelihood as certainty.
Objective
Create pipeline summary, failure investigation, rerun decision, and triage outcome for one broken build.
What to design
Define the experience
Required experience
- Identify the earliest meaningful failure
- Inspect logs, artifacts, history, and affected jobs
- Choose a scoped rerun or assign an owner
- Record the failure class and unblock or hold the change
Screens and states
- Pipeline summary
- Failure detail
- Rerun review
- Triage outcome
Core user flow
Follow the critical path
- 01
Noor opens the failed pipeline and sees downstream jobs marked not run
- 02
She reviews the timeout with recent test history
- 03
She reruns only the failed job on a fresh worker
- 04
The second failure confirms a reproducible issue and she assigns it before merging
Product rules
Requirements and constraints
Requirements
- Distinguish failed, canceled, skipped, blocked, and not-run jobs
- Surface the earliest failure with surrounding log context
- Show recent history without declaring flakiness as fact
- Estimate rerun scope, duration, and resource cost before execution
Constraints
- Logs may contain secrets and need redaction
- A rerun cannot bypass required checks
- Permissions differ between forked and internal contributions
Reality check
States worth considering
Finish line
What to deliver
- Four desktop screens covering ambiguous and confirmed failure states
Optional direction
Visual resources
Use these as a starting constraint if you want one. They are not part of the required solution.