Before: e2e \`needs: [frontend]\` waited for the frontend job to upload
a build artifact, then downloaded it. With multiple runners this means
the third runner sat idle for ~6 min while frontend ran, then started
e2e — total wall-clock max(backend, frontend+e2e) ≈ 11 min.
After: e2e builds its own frontend (npm ci + npm run build are already
in the job; just dropped the artifact download step and added the
build). e2e starts immediately on a free runner. Adds ~1-2 min to the
e2e job duration but removes ~5 min of waiting and eliminates the
cross-job artifact mechanism entirely.
Side benefit: no more \`actions/upload-artifact\` v3/v4 GHES headaches
on the cross-job handoff. The \`if: always()\` upload of the
playwright-report at the end of e2e is kept (failure report retrieval
is still useful), but it's a leaf-output, not a dependency.
Net wall-clock: max(backend=9m, frontend=6m, e2e=7m) ≈ 9 min on the
3-runner setup, down from ~11 min.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Mechanical drift between the e2e selectors and the current UI surfaced
on the first CI run after PR #149 unblocked the artifact upload step.
Five tests, three categories of drift:
1. **Page heading renames** (navigation.spec.ts)
- `Sessions` → `Session History` on /sessions
- `Account Settings` → `Account Management` on /account
2. **Route rename** (command-palette.spec.ts:74)
- The "Troubleshoot with FlowPilot" command palette option now lands
on /pilot (Phase 1 of the FlowPilot migration renamed /assistant).
/assistant still 301-redirects, so the assertion accepts either.
3. **Feature moved to /sessions** (history.spec.ts, resume.spec.ts)
- Default tab on /sessions is "AI Sessions"; flow-session filtering
and the Resume button moved behind the "Flow Sessions" tab. Both
tests now click that tab before asserting.
- resume.spec.ts no longer starts at /trees (Resume buttons aren't
rendered there anymore — the flow lives on /sessions). Destination
URL (/trees/:id/navigate) is unchanged.
No product-code changes — these are pure test updates against the
shipped UI. Run the suite locally with
`cd frontend && npm run test:e2e` once a fresh build is available.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Backend suite is the slow gate (1076 passed locally in 22m27s on
fix/ci-workflow-config). Adding pytest-xdist with per-worker DB
isolation drops it to ~4m20s on the 8-core homelab runner. Verified
locally: `pytest -n auto --no-cov` finished in 4m28s real time
(15m19s user — confirms ~5× parallelism).
How it works:
- conftest.py reads `PYTEST_XDIST_WORKER` (set per worker by xdist —
'gw0', 'gw1', …). When set, derives a per-worker DB URL like
`…/resolutionflow_test_gw0`. The base DB stays for serial / master
runs.
- `_ensure_worker_db_exists` runs synchronously at conftest import,
connects to the postgres maintenance DB, and `CREATE DATABASE`s the
worker-suffixed DB if it doesn't exist. Idempotent across runs.
- The "test" safety guard still applies — every worker DB name
contains "test" so the assertion holds.
- The per-test `DROP SCHEMA public CASCADE` now operates on the
worker's isolated DB, no cross-worker race.
CI workflow: backend job switches to `pytest -n auto`. Coverage still
collected (pytest-cov has built-in xdist support).
Adds `pytest-xdist==3.6.1` to requirements-dev.txt.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
With 3 Gitea Actions runners on the same homelab box, two simultaneous
backend (or backend + e2e) jobs both try to bind 0.0.0.0:5432 for their
postgres service containers. The second fails with:
failed to set up container networking: ... Bind for 0.0.0.0:5432
failed: port is already allocated
The host-port mapping isn't actually needed — the workflow uses
\`DATABASE_URL: postgresql+asyncpg://...@postgres:5432/...\` (hostname
\`postgres\` is the service container's docker-network DNS name).
The tests run inside the act container which is on the same docker
network, so they reach postgres without going through the host.
Removing \`ports: 5432:5432\` from both backend and e2e job service
definitions lets multiple postgres services run in parallel on
different docker networks without colliding on the host.
Surfaced when PR #150 ran in parallel with another job after the
multi-runner setup. Backend instant-failed in 2s on the docker run.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
TODO.md: Promote pytest-xdist to ✅ (PR #151 carries it). Adds three new
backlog items:
- data-testid hardening for e2e-critical interactive elements (sparked
by PR #152's selector drift work)
- per-test transactional rollback (next big speedup if needed)
- pytest-testmon for PR-time test selection
HANDOFF.md: Three open PRs now (#150, #151, #152), all independent.
Three Gitea runner agents now registered, so jobs run in parallel.
Combined with #151's xdist, the prior 1h 14m wall-clock should drop
to ~6-10 min. Updated merge order: #152 first (smallest), #150 next,
#151 last. After all three land, enable CI / backend then CI / e2e
as required status checks.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
(Was meant to land in fe632c9; the multi-line edit failed silently
because Codex's earlier entry shifted the surrounding context.)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Updates HANDOFF.md, CURRENT_TASK.md, and SESSION_LOG.md to reflect:
- PR #150 now contains the AI-provider test mock + caching + maxfail.
Backend CI should be fully green for the first time in months.
- PR #151 stacked on #150: pytest-xdist with per-worker DBs. Local
verification: 22m 27s → 4m 28s (5× speedup), 1076 passed both runs.
- DoD is now: merge #150, then #151, then add CI / backend
(pull_request) to required status checks on main.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Three changes that get PR #150 to a green CI gate:
1. **test_record_decision_persists_and_bumps_state_version** — the
`decision: draft_template` path calls `_extract_template_parameters`
(TemplateExtractionService → AI provider). CI doesn't set
ANTHROPIC_API_KEY/GOOGLE_AI_API_KEY, so the endpoint raised
`RuntimeError: No AI provider configured` and returned 500. The test
isn't exercising the AI integration — patched the extractor with an
AsyncMock returning a minimal valid `{templated_body, parameters}`
dict. Verified locally: the test now passes.
2. **pip + npm caches** in backend, frontend, and e2e jobs. Keyed on
the hash of requirements*.txt / package-lock.json with a runner-os
restore-key fallback. Saves ~30-60s per run on cache hit.
3. **Pytest invocation tightened**:
- Dropped `--cov-report=term-missing` — the custom "Display coverage
summary" step below parses coverage.json and prints the same
module list more concisely. Term-missing dumps every uncovered
line which adds ~5-10s of stdout.
- Added `--maxfail=10` so a structural breakage (fixture explosion,
DB unreachable) bails after 10 errors instead of running the full
25-min suite. Tunable.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Capture the backend pytest parallelization work so it survives session
end. Backend suite is currently ~22 min wall-clock for 1076 tests;
xdist with one-DB-per-worker should land in the 3-6 min range on the
homelab Gitea Actions runner.
Also queues two backlog items:
- Frontend lint warnings (23 react-hooks/exhaustive-deps after PR #149)
- Periodic audit of the ResourceWarning filterwarnings added by Codex
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Updates HANDOFF.md, CURRENT_TASK.md, and SESSION_LOG.md so the next
session has accurate resume state. Summary of where things are:
- PR #141 (PSA tickets), PR #147 (FlowPilot Phase 1-9), PR #148 (CI
fixes part 1), PR #149 (CI fixes part 2) all merged to main in this
session.
- Branch protection enabled on main: PR-only, CI / frontend required.
- PR #150 (this branch) is the last CI-config PR — adds
DATABASE_TEST_URL to the workflow and pins upload-artifact to v3.
- Next session: watch #150's CI, merge if green, add CI / backend to
required checks, then start on the 54 real backend test failures.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two CI-config issues blocking the gate from going green:
1. **Backend tests connect to localhost instead of postgres service.**
conftest.py reads DATABASE_TEST_URL only — DATABASE_URL is intentionally
not consulted (per dab740d's test-DB-isolation hardening — running
pytest with DATABASE_URL set previously dropped the dev DB schema).
The CI workflow only sets DATABASE_URL, so conftest falls back to its
localhost default and every fixture-setup fails with
`OSError: Connect call failed ('127.0.0.1', 5432)` — observed as 638
errors on the latest main run.
Add DATABASE_TEST_URL pointing at the postgres service container.
Same connection string as DATABASE_URL — the test DB and the app DB
are the same physical postgres in CI; conftest's safety assertion is
satisfied by the URL containing "test".
2. **Frontend artifact upload fails on Gitea Actions runner.**
actions/upload-artifact@v4 (and v5) are not supported on Gitea
Actions / GHES — the runner returns
`GHESNotSupportedError: ... not currently supported on GHES`. Lint
itself is now passing (0 errors after PR #149); the job exits 1 only
because the upload step then fails.
Pin upload-artifact + download-artifact to v3, the latest version
compatible with Gitea Actions until they ship v4 support.
After this lands, both backend and frontend CI gates should turn
green — at which point we can also add backend to the required status
checks on main.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
