libnovel/docker-compose.yml

# ── Shared environment fragments ──────────────────────────────────────────────
# These YAML anchors eliminate duplication between backend and runner.
# All values come from Doppler — no fallbacks needed here.
# Run commands via: just up / just build / etc.  (see justfile)
x-infra-env: &infra-env
  # MinIO
  MINIO_ENDPOINT: "minio:9000"
  MINIO_ACCESS_KEY: "${MINIO_ROOT_USER}"
  MINIO_SECRET_KEY: "${MINIO_ROOT_PASSWORD}"
  MINIO_USE_SSL: "false"
  MINIO_PUBLIC_ENDPOINT: "${MINIO_PUBLIC_ENDPOINT}"
  MINIO_PUBLIC_USE_SSL: "${MINIO_PUBLIC_USE_SSL}"
  # PocketBase
  POCKETBASE_URL: "http://pocketbase:8090"
  POCKETBASE_ADMIN_EMAIL: "${POCKETBASE_ADMIN_EMAIL}"
  POCKETBASE_ADMIN_PASSWORD: "${POCKETBASE_ADMIN_PASSWORD}"
  # Meilisearch
  MEILI_URL: "${MEILI_URL:-http://meilisearch:7700}"
  MEILI_API_KEY: "${MEILI_MASTER_KEY}"
  # Valkey
  VALKEY_ADDR: "valkey:6379"
  # Cloudflare AI (TTS + image generation)
  CFAI_ACCOUNT_ID: "${CFAI_ACCOUNT_ID}"
  CFAI_API_TOKEN: "${CFAI_API_TOKEN}"

services:
  # ─── MinIO (object storage: chapters, audio, avatars, browse) ────────────────
  minio:
    image: minio/minio:latest
    restart: unless-stopped
    command: server /data --console-address ":9001"
    environment:
      MINIO_ROOT_USER: "${MINIO_ROOT_USER}"
      MINIO_ROOT_PASSWORD: "${MINIO_ROOT_PASSWORD}"
    # No public port — all presigned URL traffic goes through backend or a
    # separately-exposed MINIO_PUBLIC_ENDPOINT (e.g. storage.libnovel.cc).
    expose:
      - "9000"
      - "9001"
    volumes:
      - minio_data:/data
    healthcheck:
      test: ["CMD", "mc", "ready", "local"]
      interval: 10s
      timeout: 5s
      retries: 5

  # ─── MinIO bucket initialisation ─────────────────────────────────────────────
  minio-init:
    image: minio/mc:latest
    depends_on:
      minio:
        condition: service_healthy
    entrypoint: >
      /bin/sh -c "
        mc alias set local http://minio:9000 $${MINIO_ROOT_USER} $${MINIO_ROOT_PASSWORD};
        mc mb --ignore-existing local/chapters;
        mc mb --ignore-existing local/audio;
        mc mb --ignore-existing local/avatars;
        mc mb --ignore-existing local/catalogue;
        mc mb --ignore-existing local/translations;
        mc mb --ignore-existing local/imports;
        echo 'buckets ready';
      "
    environment:
      MINIO_ROOT_USER: "${MINIO_ROOT_USER}"
      MINIO_ROOT_PASSWORD: "${MINIO_ROOT_PASSWORD}"

  # ─── PocketBase (auth + structured data) ─────────────────────────────────────
  # Custom binary built from backend/cmd/pocketbase — runs Go migrations on every
  # startup before accepting traffic, replacing the old pb-init-v3.sh script.
  pocketbase:
    image: kalekber/libnovel-pocketbase:${GIT_TAG:-latest}
    build:
      context: ./backend
      dockerfile: Dockerfile
      target: pocketbase
    labels:
      com.centurylinklabs.watchtower.enable: "true"
    restart: unless-stopped
    environment:
      # Used by migration 1 to create the initial superuser on a fresh install.
      POCKETBASE_ADMIN_EMAIL: "${POCKETBASE_ADMIN_EMAIL}"
      POCKETBASE_ADMIN_PASSWORD: "${POCKETBASE_ADMIN_PASSWORD}"
    # No public port — accessed only by backend/runner on the internal network.
    expose:
      - "8090"
    volumes:
      - pb_data:/pb_data
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://localhost:8090/api/health"]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 30s

  # ─── Meilisearch (full-text search) ──────────────────────────────────────────
  meilisearch:
    image: getmeili/meilisearch:v1.40.0
    restart: unless-stopped
    environment:
      MEILI_MASTER_KEY: "${MEILI_MASTER_KEY}"
      MEILI_ENV: "${MEILI_ENV}"
    # No public port — backend/runner reach it via internal network.
    expose:
      - "7700"
    volumes:
      - meili_data:/meili_data
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://127.0.0.1:7700/health"]
      interval: 10s
      timeout: 5s
      retries: 5

  # ─── Valkey (presign URL cache) ───────────────────────────────────────────────
  valkey:
    image: valkey/valkey:7-alpine
    restart: unless-stopped
    # No public port — backend/runner/ui reach it via internal network.
    expose:
      - "6379"
    volumes:
      - valkey_data:/data
    healthcheck:
      test: ["CMD", "valkey-cli", "ping"]
      interval: 10s
      timeout: 5s
      retries: 5

  # ─── Redis (Asynq task queue — accessed locally by backend, remotely by homelab runner) ──
  redis:
    image: redis:7-alpine
    restart: unless-stopped
    command: >
      redis-server
      --appendonly yes
      --requirepass "${REDIS_PASSWORD}"
    # No public port — backend reaches it via internal network.
    # Homelab runner reaches it via Caddy TLS proxy on :6380 → redis:6379.
    expose:
      - "6379"
    volumes:
      - redis_data:/data
    healthcheck:
      test: ["CMD", "redis-cli", "-a", "${REDIS_PASSWORD}", "ping"]
      interval: 10s
      timeout: 5s
      retries: 5

  # ─── Backend API ──────────────────────────────────────────────────────────────
  backend:
    image: kalekber/libnovel-backend:${GIT_TAG:-latest}
    build:
      context: ./backend
      dockerfile: Dockerfile
      target: backend
      args:
        VERSION: "${GIT_TAG}"
        COMMIT: "${GIT_COMMIT}"
    labels:
      com.centurylinklabs.watchtower.enable: "true"
    restart: unless-stopped
    stop_grace_period: 35s
    depends_on:
      pocketbase:
        condition: service_healthy
      minio:
        condition: service_healthy
      meilisearch:
        condition: service_healthy
      valkey:
        condition: service_healthy
      redis:
        condition: service_healthy
    # No public port — all traffic is routed via Caddy.
    expose:
      - "8080"
    environment:
      <<: *infra-env
      BACKEND_HTTP_ADDR: ":8080"
      BACKEND_ADMIN_TOKEN: "${BACKEND_ADMIN_TOKEN}"
      LOG_LEVEL: "${LOG_LEVEL}"
      KOKORO_URL: "${KOKORO_URL}"
      KOKORO_VOICE: "${KOKORO_VOICE}"
      POCKET_TTS_URL: "${POCKET_TTS_URL}"
      GLITCHTIP_DSN: "${GLITCHTIP_DSN_BACKEND}"
      OTEL_EXPORTER_OTLP_ENDPOINT: "${OTEL_EXPORTER_OTLP_ENDPOINT}"
      OTEL_SERVICE_NAME: "backend"
      # Asynq task queue — backend enqueues jobs to local Redis sidecar.
      # Homelab runner connects to the same Redis via Caddy TLS proxy on :6380.
      REDIS_ADDR: "redis:6379"
      REDIS_PASSWORD: "${REDIS_PASSWORD}"
    healthcheck:
      test: ["CMD", "/healthcheck", "http://localhost:8080/health"]
      interval: 15s
      timeout: 5s
      retries: 3

  # ─── Runner (background task worker) ─────────────────────────────────────────
  # profiles: [runner] prevents accidental restart on `docker compose up -d`.
  # The homelab runner (192.168.0.109) is the sole worker in production.
  # To start explicitly: doppler run -- docker compose --profile runner up -d runner
  runner:
    profiles: [runner]
    image: kalekber/libnovel-runner:${GIT_TAG:-latest}
    build:
      context: ./backend
      dockerfile: Dockerfile
      target: runner
      args:
        VERSION: "${GIT_TAG}"
        COMMIT: "${GIT_COMMIT}"
    labels:
      com.centurylinklabs.watchtower.enable: "true"
    restart: unless-stopped
    stop_grace_period: 135s
    depends_on:
      pocketbase:
        condition: service_healthy
      minio:
        condition: service_healthy
      meilisearch:
        condition: service_healthy
      valkey:
        condition: service_healthy
    # Metrics endpoint — internal only; expose publicly via Caddy if needed.
    expose:
      - "9091"
    environment:
      <<: *infra-env
      LOG_LEVEL: "${LOG_LEVEL}"
      # Runner tuning
      RUNNER_POLL_INTERVAL: "${RUNNER_POLL_INTERVAL}"
      RUNNER_MAX_CONCURRENT_SCRAPE: "${RUNNER_MAX_CONCURRENT_SCRAPE}"
      RUNNER_MAX_CONCURRENT_AUDIO: "${RUNNER_MAX_CONCURRENT_AUDIO}"
      RUNNER_WORKER_ID: "${RUNNER_WORKER_ID}"
      RUNNER_TIMEOUT: "${RUNNER_TIMEOUT}"
      RUNNER_METRICS_ADDR: "${RUNNER_METRICS_ADDR}"
      # Suppress the on-startup catalogue walk — catalogue_refresh now skips
      # books already in Meilisearch, so a full walk on every restart is wasteful.
      # The 24h periodic ticker (CatalogueRefreshInterval) still fires normally.
      RUNNER_SKIP_INITIAL_CATALOGUE_REFRESH: "true"
      # Kokoro-FastAPI TTS endpoint
      KOKORO_URL: "${KOKORO_URL}"
      KOKORO_VOICE: "${KOKORO_VOICE}"
      POCKET_TTS_URL: "${POCKET_TTS_URL}"
      GLITCHTIP_DSN: "${GLITCHTIP_DSN_RUNNER}"
      OTEL_EXPORTER_OTLP_ENDPOINT: "${OTEL_EXPORTER_OTLP_ENDPOINT}"
      OTEL_SERVICE_NAME: "runner"
    healthcheck:
      # 120s = 2× the default 30s poll interval with generous headroom.
      test: ["CMD", "/healthcheck", "file", "/tmp/runner.alive", "120"]
      interval: 60s
      timeout: 5s
      retries: 3

  # ─── SvelteKit UI ─────────────────────────────────────────────────────────────
  ui:
    image: kalekber/libnovel-ui:${GIT_TAG:-latest}
    build:
      context: ./ui
      dockerfile: Dockerfile
      args:
        BUILD_VERSION: "${GIT_TAG}"
        BUILD_COMMIT: "${GIT_COMMIT}"
    labels:
      com.centurylinklabs.watchtower.enable: "true"
    restart: unless-stopped
    stop_grace_period: 35s
    depends_on:
      backend:
        condition: service_healthy
      pocketbase:
        condition: service_healthy
      valkey:
        condition: service_healthy
    # No public port — all traffic via Caddy.
    expose:
      - "3000"
    environment:
      # ORIGIN must match the public URL Caddy serves on.
      # adapter-node uses this for SvelteKit's built-in CSRF origin check.
      ORIGIN: "${ORIGIN}"
      BACKEND_API_URL: "http://backend:8080"
      POCKETBASE_URL: "http://pocketbase:8090"
      POCKETBASE_ADMIN_EMAIL: "${POCKETBASE_ADMIN_EMAIL}"
      POCKETBASE_ADMIN_PASSWORD: "${POCKETBASE_ADMIN_PASSWORD}"
      AUTH_SECRET: "${AUTH_SECRET}"
      BACKEND_ADMIN_TOKEN: "${BACKEND_ADMIN_TOKEN}"
      DEBUG_LOGIN_TOKEN: "${DEBUG_LOGIN_TOKEN}"
      PUBLIC_MINIO_PUBLIC_URL: "${MINIO_PUBLIC_ENDPOINT}"
      # Valkey
      VALKEY_ADDR: "valkey:6379"
      # Umami analytics
      PUBLIC_UMAMI_WEBSITE_ID: "${PUBLIC_UMAMI_WEBSITE_ID}"
      PUBLIC_UMAMI_SCRIPT_URL: "${PUBLIC_UMAMI_SCRIPT_URL}"
      # GlitchTip client + server-side error tracking
      PUBLIC_GLITCHTIP_DSN: "${PUBLIC_GLITCHTIP_DSN}"
      # Grafana Faro RUM (browser Web Vitals, traces, errors)
      PUBLIC_FARO_COLLECTOR_URL: "${PUBLIC_FARO_COLLECTOR_URL}"
      # OpenTelemetry tracing
      OTEL_EXPORTER_OTLP_ENDPOINT: "${OTEL_EXPORTER_OTLP_ENDPOINT}"
      OTEL_SERVICE_NAME: "ui"
      # Allow large PDF/EPUB uploads (adapter-node default is 512KB)
      BODY_SIZE_LIMIT: "52428800"
      # OAuth2 providers
      GOOGLE_CLIENT_ID: "${GOOGLE_CLIENT_ID}"
      GOOGLE_CLIENT_SECRET: "${GOOGLE_CLIENT_SECRET}"
      GITHUB_CLIENT_ID: "${GITHUB_CLIENT_ID}"
      GITHUB_CLIENT_SECRET: "${GITHUB_CLIENT_SECRET}"
      # Polar (subscriptions)
      POLAR_API_TOKEN: "${POLAR_API_TOKEN}"
      POLAR_WEBHOOK_SECRET: "${POLAR_WEBHOOK_SECRET}"
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://127.0.0.1:3000/health"]
      interval: 15s
      timeout: 5s
      retries: 3

  # ─── CrowdSec (threat detection + IP blocking) ───────────────────────────────
  # Reads Caddy JSON access logs from the shared caddy_logs volume and enforces
  # decisions via the Caddy bouncer plugin.
  crowdsec:
    image: crowdsecurity/crowdsec:latest
    restart: unless-stopped
    environment:
      GID: "1000"
      COLLECTIONS: "crowdsecurity/caddy crowdsecurity/http-dos crowdsecurity/base-http-scenarios"
    volumes:
      - crowdsec_data:/var/lib/crowdsec/data
      - ./crowdsec/acquis.yaml:/etc/crowdsec/acquis.yaml:ro
      - caddy_logs:/var/log/caddy:ro
    expose:
      - "8080"
    healthcheck:
      test: ["CMD", "cscli", "version"]
      interval: 20s
      timeout: 10s
      retries: 5

  # ─── Dozzle agent ────────────────────────────────────────────────────────────
  # Exposes prod container logs to the Dozzle instance on the homelab.
  # The homelab Dozzle connects here via DOZZLE_REMOTE_AGENT.
  # Port 7007 is bound to localhost only — not reachable from the internet.
  dozzle-agent:
    image: amir20/dozzle:latest
    restart: unless-stopped
    command: agent
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
    ports:
      - "127.0.0.1:7007:7007"

  # ─── CrowdSec bouncer registration ───────────────────────────────────────────
  # One-shot: registers the Caddy bouncer with the CrowdSec LAPI and writes the
  # generated API key to crowdsec/.crowdsec.env, which Caddy reads via env_file.
  # Uses the Docker socket to exec cscli inside the running crowdsec container.
  crowdsec-init:
    image: docker:cli
    depends_on:
      crowdsec:
        condition: service_healthy
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./crowdsec:/crowdsec-out
    entrypoint: >
      /bin/sh -c "
        out=/crowdsec-out/.crowdsec.env;
        existing=$$(grep -s '^CROWDSEC_API_KEY=.' \"$$out\" | cut -d= -f2-);
        if [ -n \"$$existing\" ]; then
          echo 'crowdsec-init: key already present, skipping registration';
          exit 0;
        fi;
        container=$$(docker ps --filter name=crowdsec --filter status=running --format '{{.Names}}' | grep -v init | head -1);
        echo \"crowdsec-init: using container $$container\";
        docker exec $$container cscli bouncers delete caddy-bouncer 2>/dev/null || true;
        key=$$(docker exec $$container cscli bouncers add caddy-bouncer -o raw 2>&1);
        if [ -z \"$$key\" ]; then
          echo 'crowdsec-init: ERROR — failed to obtain bouncer key' >&2;
          exit 1;
        fi;
        printf 'CROWDSEC_API_KEY=%s\n' \"$$key\" > \"$$out\";
        echo \"crowdsec-init: bouncer key written (key length: $${#key})\";
      "
    restart: "no"


  # ─── Caddy (reverse proxy + automatic HTTPS) ──────────────────────────────────
  # Custom build includes github.com/mholt/caddy-ratelimit,
  # github.com/hslatman/caddy-crowdsec-bouncer/http, and
  # github.com/mholt/caddy-l4 (TCP layer4 proxy for Redis).
  caddy:
    image: kalekber/libnovel-caddy:${GIT_TAG:-latest}
    build:
      context: ./caddy
      dockerfile: Dockerfile
    labels:
      com.centurylinklabs.watchtower.enable: "true"
    restart: unless-stopped
    depends_on:
      backend:
        condition: service_healthy
      ui:
        condition: service_healthy
      crowdsec-init:
        condition: service_completed_successfully
    ports:
      - "80:80"
      - "443:443"
      - "443:443/udp" # HTTP/3 (QUIC)
      - "6380:6380"   # Redis TCP proxy (TLS) for homelab runner → Asynq
    environment:
      DOMAIN: "${DOMAIN}"
      CADDY_ACME_EMAIL: "${CADDY_ACME_EMAIL}"
    env_file:
      - path: ./crowdsec/.crowdsec.env
        required: false
    volumes:
      - ./Caddyfile:/etc/caddy/Caddyfile:ro
      - ./caddy/errors:/srv/errors:ro
      - caddy_data:/data
      - caddy_config:/config
      - caddy_logs:/var/log/caddy

  # ─── Watchtower (auto-redeploy custom services on new images) ────────────────
  # Only watches services labelled com.centurylinklabs.watchtower.enable=true.
  # Third-party infra images (minio, pocketbase, meilisearch, etc.) are excluded.
  # doppler binary is mounted from the host so watchtower fetches fresh secrets
  # on every start (notification URL, credentials) without baking them in.
  watchtower:
    image: containrrr/watchtower:latest
    restart: unless-stopped
    entrypoint: ["/usr/bin/doppler", "run", "--project", "libnovel", "--config", "prd", "--"]
    command: ["/watchtower", "--label-enable", "--interval", "300", "--cleanup"]
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - /usr/bin/doppler:/usr/bin/doppler:ro
      - /root/.doppler:/root/.doppler:ro
    environment:
      HOME: "/root"
      DOCKER_API_VERSION: "1.44"

volumes:
  minio_data:
  pb_data:
  meili_data:
  valkey_data:
  redis_data:
  caddy_data:
  caddy_config:
  caddy_logs:
  crowdsec_data: