feat(repo): dockerize full stack (#275)

[w7-mgfcode]

Summary

Implements PRP-32 — packages ForecastLabAI as a four-service Docker Compose stack so make docker-up brings the whole system (Postgres + backend + frontend + opt-in Ollama) up on any host that has Docker, with a separate GPU override for Ollama NVIDIA passthrough.

Closes #275.

Changes

• Dockerfile.backend — multi-stage python:3.12-slim-bookworm + uv 0.11.8 (base → deps → dev → prod). Asyncpg-only path; no libpq.
• Dockerfile.frontend — multi-stage Node 22 + pnpm via corepack (builder → prod nginx:1.27-alpine → dev vite). PNPM_CONFIG_DANGEROUSLY_ALLOW_ALL_BUILDS=true works around pnpm 11 ERR_PNPM_IGNORED_BUILDS.
• .dockerignore — excludes .venv, node_modules, docs/, examples/, PRPs/. Keeps README.md (uv reads it via pyproject.toml.readme during uv sync).
• docker-compose.yml — adds backend, frontend, opt-in ollama services on a forecastlab bridge network; backend overrides DATABASE_URL → postgres:5432 and OLLAMA_BASE_URL → http://ollama:11434; frontend keeps VITE_* URLs on localhost:8123 (browser-is-consumer rule).
• docker-compose.gpu.yml — Ollama NVIDIA passthrough override.
• Makefile — docker-up, docker-up-gpu, docker-down targets.
• Docs — README.md quick-start subsection, docs/_base/RUNBOOKS.md H3 with 7 failure modes, docs/rag-ollama-setup.md in-cluster-vs-host URL paragraph (EN/HU).
• Tests — tests/test_docker_stack.py: 3 integration tests gated on DOCKER_STACK_TEST=1 (services_healthy, backend↔postgres internal DNS, no-hardcoded-localhost static check).
• uv.lock — incidental 0.2.16 → 0.2.17 bump (split into its own chore(repo) commit).

Validation (executed locally before PR open)

• L1 — docker compose config (base + gpu) clean; ruff check . + ruff format --check . clean.
• L2 — mypy app/ 0 errors (276 files); pyright app/ 0 errors; pytest -m "not integration" 1430 passed.
• L3 — make docker-up brought all 3 services Healthy in ~30 s; DOCKER_STACK_TEST=1 pytest -m integration tests/test_docker_stack.py 3/3 passed.
• L4 — /health, /docs, SPA index, backend ↔ postgres:5432 via internal DNS all green; make docker-down clean teardown.
• L5 (GPU) — skipped (host lacks nvidia-container-runtime, per PRP "SKIP on non-GPU CI").

Risk + rollback

• Packaging-only change — no pyproject.toml / package.json dependency additions; no schema migrations; no API surface change.
• Host-mode unchanged — existing docker compose up -d postgres + host-side uvicorn flow keeps working; the new services are additive.
• Rollback — revert the 4 commits on this branch; nothing in dev is altered destructively.

Test plan

• CI: Lint & Format, Type Check, Test, Migration Check all green.
• Reviewer can run make docker-up and hit http://localhost:8123/health + http://localhost:5173/.
• Reviewer can run make docker-down for clean teardown.

[sourcery-ai]

Sorry @w7-mgfcode, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: ec5d90bc-d2d3-4a59-85fe-9b728ffafcd8

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/repo-dockerize-full-stack

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}