gendesign/docker-compose.prod.yml
lekss361 0c48166af6 chore(infra): correct frontend healthcheck + tighten depends_on (re-apply #117)
Forward-fix после incident May 14 (см. vault events/Incident_Prod_Down_May14_2026).
PR #117 был полностью reverted PR #119 (Caddy depends_on frontend: service_healthy
+ failing wget --spider /  → Caddy не стартовал → vsё прод down ~25 мин).

Этот PR делает то же что планировал #117, но корректно:

### Frontend healthcheck — TCP probe через node (НЕ HTTP)

```yaml
test: ["CMD", "node", "-e",
  "require('net').createConnection({port:3000,host:'127.0.0.1'})
   .once('connect',function(){process.exit(0)})
   .once('error',function(){process.exit(1)})"]
start_period: 60s
retries: 6
interval: 15s
```

- НЕ зависит от HTTP status (200/404/308 — все OK если порт слушает)
- `node` гарантированно в alpine image (запускает server.js)
- `127.0.0.1` — никакой DNS ambiguity (busybox alpine localhost IPv6 vs IPv4)
- 60s + 6×15s = 150s window — с запасом на Next.js standalone cold start

### Caddy depends_on — backend healthy, frontend started (НЕ healthy)

```yaml
caddy:
  depends_on:
    backend: { condition: service_healthy }    # backend имеет /health endpoint
    frontend: { condition: service_started }   # safety net — НЕ блокирует Caddy
```

Lesson from incident: frontend health flaky (Next.js cold start variable).
Strict service_healthy для frontend → одна misconfig = full outage обоих доменов
(gendsgn.ru + obsidian.gendsgn.ru через тот же Caddy на shared network).

### backend/worker/beat redis: service_started → service_healthy

Redis уже имеет `redis-cli ping` healthcheck. Tightening безопасно — был
artifact pre-healthcheck Redis.

### Pin rosreestr2coord>=5.0.0 (lock уже 5.3.3)

v5 убрала delay параметр; код адаптирован (rate limit на worker уровне).
Защита от silent downgrade при `uv lock --upgrade`.

### Что НЕ повторено из #117

- Caddy depends_on frontend: service_healthy — заменено на service_started
- wget --spider / — заменено на node TCP probe

### Vault обновлён

- `events/Incident_Prod_Down_May14_2026.md` — постмортем + timeline + lessons
- Update Changelog в `domains/infra/infra-MOC.md` (если есть)

Closes incident. Re-applies #117 goals safely.
2026-05-14 23:15:46 +03:00

161 lines
5.9 KiB
YAML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Production compose. Pulls pre-built images from GHCR
# (built and pushed by .github/workflows/deploy.yml).
#
# Place at /opt/gendesign/docker-compose.prod.yml on the VM.
# Required env in /opt/gendesign/.env:
# POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD
# IMAGE_TAG (defaults to "latest")
#
# In production the browser talks to Caddy on :80/:443.
# Caddy proxies /api/* and /health straight to backend, the rest to frontend.
# So the frontend uses same-origin relative URLs — no NEXT_PUBLIC_API_BASE_URL needed.
#
# Postgres + Redis run alongside the app on the same VM (Discovery mode).
# Volumes are shared with docker-compose.yml so switching between files preserves data.
services:
postgres:
image: postgis/postgis:16-3.4
restart: unless-stopped
environment:
POSTGRES_DB: ${POSTGRES_DB}
POSTGRES_USER: ${POSTGRES_USER}
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
# Bind to 127.0.0.1 only — accessible from host (for SSH tunnel) but not from public internet.
# UFW additionally blocks 5432 from outside.
ports:
- "127.0.0.1:5432:5432"
volumes:
- postgres_data:/var/lib/postgresql/data
- ./backend/db/init:/docker-entrypoint-initdb.d:ro
healthcheck:
test: ["CMD-SHELL", "pg_isready -U $${POSTGRES_USER}"]
interval: 5s
timeout: 5s
retries: 10
redis:
image: redis:7-alpine
restart: unless-stopped
volumes:
- redis_data:/data
healthcheck:
test: ["CMD", "redis-cli", "ping"]
interval: 10s
timeout: 3s
retries: 5
backend:
image: ghcr.io/lekss361/gendesign-backend:${IMAGE_TAG:-latest}
restart: unless-stopped
# .env.runtime пишется deploy.yml через SSH (SENTRY_RELEASE=$IMAGE_TAG).
# required: false — compose не падает если файла нет (первый деплой).
env_file:
- path: ./backend/.env
- path: ./backend/.env.runtime
required: false
depends_on:
postgres:
condition: service_healthy
redis:
condition: service_healthy
ports:
- "127.0.0.1:8000:8000"
healthcheck:
test: ["CMD", "curl", "-fsS", "http://localhost:8000/health"]
interval: 30s
timeout: 5s
retries: 5
start_period: 20s
frontend:
image: ghcr.io/lekss361/gendesign-frontend:${IMAGE_TAG:-latest}
restart: unless-stopped
ports:
- "127.0.0.1:3000:3000"
depends_on:
- backend
# TCP-only probe через node (есть в alpine image, запускает сам server.js).
# НЕ использует HTTP semantics — проверяет только что порт 3000 слушает.
# Если Next.js процесс жив и порт открыт → healthy, независимо от HTTP status.
# start_period 60s даёт время Next.js standalone bundle bootstrap.
healthcheck:
test: ["CMD", "node", "-e", "require('net').createConnection({port:3000,host:'127.0.0.1'}).once('connect',function(){process.exit(0)}).once('error',function(){process.exit(1)})"]
interval: 15s
timeout: 5s
retries: 6
start_period: 60s
worker:
# Отдельный chromium-образ (+200 МБ Playwright). См. backend/Dockerfile target=runner-with-chromium.
image: ghcr.io/lekss361/gendesign-worker:${IMAGE_TAG:-latest}
restart: unless-stopped
env_file:
- path: ./backend/.env
- path: ./backend/.env.runtime
required: false
depends_on:
postgres:
condition: service_healthy
redis:
condition: service_healthy
volumes:
- ./data:/app/data # playwright_state.json + photo binaries
# Read-only bind-mount Антоновского /sf/ SQLite — для objective_etl task.
# На VPS файл лежит в /opt/gendesign/site-finder/analysis.db; путь внутри
# контейнера задан settings.objective_anton_sqlite_path (default
# /data/anton-sqlite/analysis.db).
- /opt/gendesign/site-finder:/data/anton-sqlite:ro
command: ["celery", "-A", "app.workers.celery_app", "worker", "--loglevel=info", "--concurrency=8", "--queues=celery,scrape_kn,geo"]
beat:
# Lean backend-образ (без Chromium) — beat только триггерит таски в Redis.
image: ghcr.io/lekss361/gendesign-backend:${IMAGE_TAG:-latest}
restart: unless-stopped
env_file:
- path: ./backend/.env
- path: ./backend/.env.runtime
required: false
depends_on:
redis:
condition: service_healthy
command: ["celery", "-A", "app.workers.celery_app", "beat", "--loglevel=info"]
caddy:
image: caddy:2
restart: unless-stopped
ports:
- "80:80"
- "443:443"
volumes:
- ./Caddyfile:/etc/caddy/Caddyfile:ro
- caddy_data:/data
- caddy_config:/config
# backend: service_healthy — backend имеет /health endpoint, готов до Caddy.
# frontend: service_started — НЕ service_healthy: даже если frontend healthcheck
# дребезжит, Caddy всё равно стартует (можно отдавать 502 для /, но
# obsidian.gendsgn.ru, api.gendsgn.ru и т.д. остаются доступны).
depends_on:
backend:
condition: service_healthy
frontend:
condition: service_started
networks:
- default
# shared — для маршрута obsidian.gendsgn.ru → couchdb (отдельный stack).
# Если obsidian-стек не задеплоен, Caddy просто получит 502 на этом маршруте,
# main-приложение не страдает.
- shared
volumes:
postgres_data:
redis_data:
caddy_data:
caddy_config:
networks:
# Внешняя сеть, создаётся вне compose (см. docs/obsidian-livesync.md).
# Связывает main-stack (Caddy) и obsidian-stack (CouchDB).
shared:
external: true
name: gendesign_shared