ubiquitous-invention/docker/docker-compose.coolify.yml

# Docker Compose for Coolify deployment of ECHODO.
#
# Usage in Coolify:
#   - New Resource → Docker Compose → connect this repo
#   - Build pack: Docker Compose
#   - Compose file path: docker/docker-compose.coolify.yml
#   - Set domains in Coolify on the `web` and `collab` services using http://
#     (CT 100 Traefik handles TLS termination per AGENT-DEPLOY.md)
#
# Path conventions:
#   Coolify invokes `docker compose` with --project-directory set to the
#   repository root, NOT this file's directory. So all build contexts and
#   dockerfile paths below are written relative to the repo root.
#
# Differences vs. docker/docker-compose.yml:
#   - No bundled postgres / redis (uses shared CT 102 services)
#   - No host port mappings (Coolify's Traefik routes by container labels)
#   - Uses Coolify SERVICE_FQDN_* magic env vars so Coolify auto-wires Traefik labels
#
# Build strategy:
#   All three services share docker/Dockerfile and select runtime via `target:`.
#   The shared `deps` stage is hashed identically across targets, so BuildKit
#   runs `pnpm install` ONCE.
#
#   The heavy `next build` for `web` (peak ~3 GB heap with 5 GB headroom)
#   must not race the two tsup builds for collab+mcp on the 8 GB Coolify
#   host or the kernel OOM-kills it (exit 255 with no Docker error). Build
#   ordering is enforced inside docker/Dockerfile: the `web-build` stage
#   has explicit `COPY --from=collab-build` / `COPY --from=mcp-build` lines
#   that force BuildKit's DAG to finish those two before starting the web
#   compile. The `depends_on` below only orders runtime startup, NOT image
#   builds — that was a misconception from an earlier iteration of this
#   file.

services:
  web:
    build:
      context: .
      dockerfile: docker/Dockerfile
      target: web
    depends_on:
      - collab
      - mcp
    restart: unless-stopped
    environment:
      # Coolify FQDN magic var — set the domain in the Coolify UI for this service.
      # SERVICE_FQDN_WEB_3000 makes Coolify expose the container on port 3000 via Traefik.
      - SERVICE_FQDN_WEB_3000
      - DATABASE_URL=${DATABASE_URL}
      - REDIS_URL=${REDIS_URL}
      - NEXTAUTH_URL=${NEXTAUTH_URL}
      - NEXTAUTH_SECRET=${NEXTAUTH_SECRET}
      - AUTH_SECRET=${AUTH_SECRET}
      - AUTH_DEV_PASSWORD=${AUTH_DEV_PASSWORD}
      - AUTH_GITHUB_ID=${AUTH_GITHUB_ID}
      - AUTH_GITHUB_SECRET=${AUTH_GITHUB_SECRET}
      - AUTH_GOOGLE_ID=${AUTH_GOOGLE_ID}
      - AUTH_GOOGLE_SECRET=${AUTH_GOOGLE_SECRET}
      - AUTH_AUTHENTIK_ID=${AUTH_AUTHENTIK_ID}
      - AUTH_AUTHENTIK_SECRET=${AUTH_AUTHENTIK_SECRET}
      - AUTH_AUTHENTIK_ISSUER=${AUTH_AUTHENTIK_ISSUER}
      - COLLAB_SERVER_URL=${COLLAB_SERVER_URL}
      - NEXT_PUBLIC_COLLAB_SERVER_URL=${NEXT_PUBLIC_COLLAB_SERVER_URL}
      - NEXT_PUBLIC_UMAMI_SCRIPT=${NEXT_PUBLIC_UMAMI_SCRIPT}
      - NEXT_PUBLIC_UMAMI_WEBSITE_ID=${NEXT_PUBLIC_UMAMI_WEBSITE_ID}
      - NEXT_PUBLIC_LIBREDESK_URL=${NEXT_PUBLIC_LIBREDESK_URL}
      - NEXT_PUBLIC_LIBREDESK_WIDGET_ID=${NEXT_PUBLIC_LIBREDESK_WIDGET_ID}
      - NEXT_PUBLIC_DIRECTUS_URL=${NEXT_PUBLIC_DIRECTUS_URL}
      - DIRECTUS_FEEDBACK_TOKEN=${DIRECTUS_FEEDBACK_TOKEN}
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - OPENAI_BASE_URL=${OPENAI_BASE_URL}
      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
      - NODE_ENV=production
      - NEXT_TELEMETRY_DISABLED=1
    expose:
      - "3000"
    healthcheck:
      test:
        - CMD-SHELL
        - "wget -qO- --timeout=3 http://127.0.0.1:3000/ >/dev/null || exit 1"
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 90s
    networks:
      - echodo

  collab:
    build:
      context: .
      dockerfile: docker/Dockerfile
      target: collab
    restart: unless-stopped
    environment:
      # Separate FQDN for the websocket service, e.g. collab.echodo.stillwell.cloud
      - SERVICE_FQDN_COLLAB_1234
      - DATABASE_URL=${DATABASE_URL}
      - REDIS_URL=${REDIS_URL}
      - PORT=1234
      - NODE_ENV=production
    expose:
      - "1234"
    healthcheck:
      test:
        - CMD-SHELL
        - 'node -e "const p=+(process.env.PORT||1234);require(''net'').createConnection(p,''127.0.0.1'').on(''connect'',()=>process.exit(0)).on(''error'',()=>process.exit(1))"'
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 25s
    networks:
      - echodo

  # MCP server is stdio-based today; deploy only if you want it reachable on the LAN
  # for future HTTP/SSE transport. Comment the whole block out if not needed.
  mcp:
    build:
      context: .
      dockerfile: docker/Dockerfile
      target: mcp
      args:
        MCP_SERVER_PORT: ${MCP_SERVER_PORT:-3001}
    restart: unless-stopped
    environment:
      - DATABASE_URL=${DATABASE_URL}
      - MCP_SERVER_PORT=${MCP_SERVER_PORT:-3001}
      - NODE_ENV=production
    stdin_open: true
    expose:
      - "${MCP_SERVER_PORT:-3001}"
    healthcheck:
      test:
        - CMD-SHELL
        - "pgrep -f 'node.*dist/index\\.js' > /dev/null || exit 1"
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 25s
    networks:
      - echodo

networks:
  echodo:
    driver: bridge