enviodev/ci

Investigating CI Failures

This repo is public, so unauthenticated GitHub API calls work. Use gh if available, otherwise fall back to curl + python3.

Step 1: Find the latest run and its jobs

branch=$(git rev-parse --abbrev-ref HEAD) && \
curl -sf "https://api.github.com/repos/enviodev/hyperindex/actions/runs?branch=${branch}&per_page=5" \
  | python3 -c "
import sys, json
runs = json.load(sys.stdin)['workflow_runs']
if not runs:
    print('No runs found'); sys.exit(0)
for r in runs[:5]:
    status = 'FAIL' if r['conclusion'] == 'failure' else 'PASS' if r['conclusion'] == 'success' else r['status']
    print(f'[{status}]  {r[\"id\"]}  {r[\"name\"]}  {r[\"created_at\"]}')"

Then list jobs for the failed run:

curl -sf "https://api.github.com/repos/enviodev/hyperindex/actions/runs/RUN_ID/jobs" \
  | python3 -c "
import sys, json
jobs = json.load(sys.stdin)['jobs']
for j in jobs:
    mark = 'FAIL' if j['conclusion'] == 'failure' else 'ok'
    print(f'[{mark}] {j[\"name\"]} (job id: {j[\"id\"]})')
    if j['conclusion'] == 'failure':
        for s in j['steps']:
            if s['conclusion'] == 'failure':
                print(f'       step failed: {s[\"name\"]}')"

Step 2: Get failed job logs

Use gh if available (best output), otherwise use WebFetch on the job URL:

# Option A: gh (if available)
gh run view RUN_ID --log-failed

# Option B: WebFetch the job page for a summary
# https://github.com/enviodev/hyperindex/actions/runs/RUN_ID/job/JOB_ID

The output can be large. Focus on the last 200 lines per job, or grep for error, FAILED, assert, or panic.

Step 3: Check annotations (supplementary)

Annotations give file:line locations when available (e.g., lint errors, compiler errors), but many failures (tests, build) produce no annotations:

curl -sf "https://api.github.com/repos/enviodev/hyperindex/check-runs/JOB_ID/annotations" \
  | python3 -c "
import sys, json
anns = json.load(sys.stdin)
if not anns:
    print('No annotations found')
else:
    for a in anns:
        print(f'{a[\"path\"]}:{a[\"start_line\"]} - {a[\"message\"][:300]}')"

Diagnosis checklist

• Read the error message carefully — most CI failures are test failures or build errors
• Check if the failure is flaky by looking at the same workflow on the base branch
• If logs are truncated and gh is available, re-run with gh run rerun RUN_ID --failed