// Debug e2e test failures, triage CI failures, analyze Cypress test logs, investigate flaky tests, diagnose test failures, check PR test results, why did tests fail, debug cypress, analyze test artifacts
Create a Jira ticket, file a bug, open a Jira issue, draft a story, log a spike, raise an issue, create a task, report a bug, write a ticket, submit a Jira, new Jira ticket, open a task, file an issue, draft a bug report, create Jira bug, create Jira task
Create a pull request, open a PR, submit a PR, push and create PR, send for review, open pull request, make a PR, raise a pull request
| name | debug-e2e-tests |
| description | Debug e2e test failures, triage CI failures, analyze Cypress test logs, investigate flaky tests, diagnose test failures, check PR test results, why did tests fail, debug cypress, analyze test artifacts |
Investigate failed Cypress e2e test runs by downloading logs and artifacts from GitHub Actions, analyzing test failures, and suggesting fixes.
gh repo view --json nameWithOwner -q .nameWithOwner
Expected repo: konflux-ci/konflux-ui
The user provides a run URL, run ID, PR number, or asks about "the latest failure".
HEAD_SHA=$(gh pr view PR_NUMBER --repo konflux-ci/konflux-ui --json headRefOid -q .headRefOid)
gh api "repos/konflux-ci/konflux-ui/actions/runs?head_sha=$HEAD_SHA&per_page=5" \
--jq '.workflow_runs[] | {id, name, status, conclusion}'
gh run list --repo konflux-ci/konflux-ui --workflow "PR Check Test" --status failure --limit 5
Single command — gets run status, downloads logs artifact, and lists files:
ARTIFACT_DIR=$(mktemp -d) && echo "ARTIFACT_DIR=$ARTIFACT_DIR" && \
gh run view RUN_ID --repo konflux-ci/konflux-ui --json status,conclusion,jobs \
--jq '{status: .status, conclusion: .conclusion, failed_jobs: [.jobs[] | select(.conclusion == "failure") | .name], all_jobs: [.jobs[] | {name: .name, status: .status, conclusion: .conclusion}]}' && \
echo "---" && \
gh run download RUN_ID --repo konflux-ci/konflux-ui --name logs --dir "$ARTIFACT_DIR" 2>&1 && \
echo "--- artifacts:" && \
find "$ARTIFACT_DIR" -type f | sort
If status is in_progress, artifacts may still exist from a completed test step — proceed with analysis.
If download fails (no artifacts), fall back to: gh run view RUN_ID --repo konflux-ci/konflux-ui --log-failed
Use the artifact file list from Step 2. Analyze in priority order — stop early if root cause is clear.
File paths: The find output from Step 2 prints absolute paths — use those directly when reading files.
Cascade awareness: The test suite runs with testIsolation: false — tests within a describe block share state. Find the first failure chronologically; subsequent failures in the same block are likely cascades. Focus on the root failure only.
junit-*.xml) — Which tests failed, error messages, durations. Identify cascades here.cypress-log.txt) — Full terminal output. Search for CypressError, AssertionError, or failing to locate the error, then read the 10-20 commands before it for context.screenshots/*.png) — Visual state at failure. Use the Read tool.saved-doms/*.html) — Check for: element not visible, error banner/modal blocking UI, loading state, logged out.| File | What to look for |
|---|---|
konflux-ui.log | Proxy pod logs, API routing errors |
pipelinerun-res.log | PipelineRun status (empty = none created) |
taskrun-res.log | TaskRun status, Tekton-level failures |
failed-pods-logs.log | Container failures, OOMKilled, CrashLoopBackOff (only if Warning events exist) |
failed-pods-definitions.yaml | Pod specs for failed pods (only if Warning events exist) |
operator-logs.log | Operator crashes, reconciliation errors |
system-resources.log | Node CPU/memory pressure, disk usage |
cluster-resources.log | Cluster-wide resource overview |
container-resources.log | Per-container resource usage |
mem.log | Memory usage over time during test run |
konflux-crs-status.log | Custom Resource readiness conditions |
failed-deployment-event-log.log | Deployment rollout failures |
kyverno-policy-pods.log | Kyverno policy violations |
artifacts/) — events.json for timing issues, pipelineruns.json, components.json. Pod logs in artifacts/pods/ (especially build-service and PaC controller).index.html (mochawesome — too large to read) and videos/*.mp4 (can't view in CLI).Check what code was changed: gh pr diff PR_NUMBER --repo konflux-ci/konflux-ui
Cross-reference: selector changes (data-test), API endpoint changes, component modifications.
| Category | Description | Where to fix |
|---|---|---|
| A. Test code issue | Bad selector, missing wait, wrong assertion | e2e-tests/ |
| B. UI bug | Component rendering error, broken user flow | src/ |
| C. Infrastructure flake | WebSocket drop, cluster timeout, resource exhaustion | Test resilience |
| D. Pipeline/backend | Build failure, API error, auth problem | Backend/config |
| E. Cascade failure | Caused by an earlier test failing | Fix root failure |
## E2E Test Failure Analysis
**Run:** [link to GH Actions run]
**PR:** [link to PR, if applicable]
**Failed tests:** X of Y (Z cascade failures)
### Root Failure
**Test:** [test name]
**Category:** [A/B/C/D/E] — [category name]
**Error:** [one-line error message]
**Analysis:**
[2-3 sentences explaining what happened and why]
**Suggested fix:**
[Concrete suggestion — file path, what to change, or "retry if flake"]
### Cascade Failures (if any)
- [test name] — caused by root failure above
### Artifacts reviewed
- [list of files that were analyzed]
| Pattern | Likely cause | Quick check |
|---|---|---|
Timed out retrying: cy.get() - element not found | Selector changed or element not rendered | Check DOM snapshot for element presence |
expected X to equal Y but values look correct | Whitespace or timing issue | Check if text has leading/trailing spaces |
cy.get() - exceeded timeout of 40000ms | Slow page load or missing element | Check infrastructure logs for API errors |
Status still running. Reloading page (0 retries remaining) | Pipeline too slow or websocket dropped | Check system-resources.log for resource issues |
| All tests after a certain point failed | Cascade failure | Find the first failure |
| Login-related errors in before() hook | Auth/cluster issue | Check infrastructure logs for 401/403 |