argus-installer/docker-compose.yml

services:
  # TimescaleDB (PostgreSQL with time-series optimization)
  db:
    image: timescale/timescaledb:latest-pg17
    container_name: argus-db
    restart: unless-stopped
    logging:
      driver: "json-file"
      options:
        max-size: "50m"
        max-file: "3"
    environment:
      TZ: ${TZ:-UTC}
      # Internal database credentials, not customer-configurable — nothing
      # outside this compose stack ever connects to Postgres directly, so
      # there's no real reason to let these vary. Only the password (below)
      # needs to be a real secret; the username/db name are just labels.
      POSTGRES_USER: postgres
      POSTGRES_PASSWORD: ${DB_PASSWORD:-postgres}
      POSTGRES_DB: argus
    command: postgres -c timezone=${TZ:-UTC} -c log_timezone=${TZ:-UTC} -c shared_preload_libraries=timescaledb -c timescaledb.telemetry_level=off -c max_locks_per_transaction=512
    volumes:
      - postgres_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U postgres"]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 30s

  # Valkey (Redis-compatible) for caching
  valkey:
    image: valkey/valkey:9-alpine
    container_name: argus-valkey
    restart: unless-stopped
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
    environment:
      TZ: ${TZ:-UTC}
    command: >
      valkey-server
      --maxmemory 256mb
      --maxmemory-policy allkeys-lru
      --appendonly yes
      --appendfsync everysec
    volumes:
      - valkey_data:/data
    healthcheck:
      test: ["CMD", "valkey-cli", "ping"]
      interval: 10s
      timeout: 5s
      retries: 5
      start_period: 10s

  # Restricts the api service's Docker Engine API access to exactly what it
  # needs (list/inspect/exec/restart existing containers for log collection
  # and nginx reload) — it cannot create new containers or touch the host
  # beyond that. See docker-socket-proxy/README.md for the exact allowed set.
  docker-socket-proxy:
    image: tecnativa/docker-socket-proxy:latest
    container_name: argus-docker-socket-proxy
    restart: unless-stopped
    security_opt:
      - "no-new-privileges:true"
    environment:
      CONTAINERS: 1
      IMAGES: 1
      VOLUMES: 1
      SYSTEM: 1
      EXEC: 1
      VERSION: 1
      POST: 1
      # Restart (stop/restart/kill on an *existing* container) only --
      # create/prune stay denied by the haproxy template's own explicit
      # deny-list regardless of POST, and start/stop stay off (ALLOW_START/
      # ALLOW_STOP unset) since nothing needs them yet. This comment block's
      # own claim of "restart" capability (above) was aspirational until
      # this line existed -- appliance#43's Diagnostics recovery actions are
      # the first real caller.
      ALLOW_RESTARTS: 1
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - ./docker-socket-proxy/haproxy.cfg.template:/usr/local/etc/haproxy/haproxy.cfg.template:ro
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"

  # Go API Server
  api:
    # ARGUS_VERSION is set by update.sh before `docker compose pull`, pinned
    # to a specific already-fully-promoted version tag rather than the
    # floating :latest (ADR-0051 -- argus-appliance repo). Manual/non-
    # update.sh use (plain `docker compose pull`, no env var set) is
    # unaffected: the default is :latest, identical to before this change.
    image: git-cloud.weboria.eu/weboria/argus-api:${ARGUS_VERSION:-latest}
    container_name: argus-api
    restart: unless-stopped
    # Container hardening: no-new-privileges blocks setuid privilege
    # escalation; cap_drop ALL removes every Linux capability, re-adding only
    # what's needed (DAC_OVERRIDE for the shared /etc/nginx volume owned by
    # the nginx user, NET_BIND_SERVICE for the DNS Security Engine's :53).
    security_opt:
      - "no-new-privileges:true"
    cap_drop:
      - ALL
    cap_add:
      - DAC_OVERRIDE
      - NET_BIND_SERVICE
    logging:
      driver: "json-file"
      options:
        max-size: "100m"
        max-file: "5"
    extra_hosts:
      - "host.docker.internal:host-gateway"
    environment:
      TZ: ${TZ:-UTC}
      PORT: "8080"
      # DATABASE_URL is not set here — the API builds it internally from
      # DB_PASSWORD using net/url (proper percent-encoding), instead of this
      # file naively interpolating the password into a raw connection
      # string. A raw '${DB_PASSWORD}@...' interpolation breaks outright if
      # the password ever contains '@', ':', '/', or similar (confirmed
      # live: an openssl-rand-base64-generated password containing '/'
      # produced an unparseable URL and the API never became healthy).
      DB_PASSWORD: ${DB_PASSWORD:-postgres}
      REDIS_URL: redis://valkey:6379/0
      ENVIRONMENT: ${ENVIRONMENT:-production}
      NGINX_CONTAINER: argus-proxy
      NGINX_SKIP_TEST: "false"
      NGINX_STATUS_URL: "http://host.docker.internal:${NGINX_HTTP_PORT:-80}/nginx_status"
      NGINX_ACCESS_LOG: "/etc/nginx/logs/access_raw.log"
      BACKUP_PATH: ${BACKUP_PATH:-/app/data/backups}
      DOCKER_API_VERSION: ${DOCKER_API_VERSION:-}
      DOCKER_HOST: tcp://docker-socket-proxy:2375
      NGINX_HTTP_PORT: ${NGINX_HTTP_PORT:-}
      NGINX_HTTPS_PORT: ${NGINX_HTTPS_PORT:-}
      API_HOST_PORT: ${API_HOST_PORT:-9080}
      API_HOST: ${API_HOST:-}
      # ANIS community intelligence — on by default against the public
      # Community Edition hub (community tier, no signup required). There is
      # no self-hosted-ANIS deployment model: the URL is fixed, not
      # customer-configurable. A license key upgrades the tier, but is only
      # ever set from Threat Intel -> ANIS Connection in the dashboard
      # (write-only there — never round-trips back to the client), not from
      # this file, so it's seeded empty (community tier) here.
      # (No ANIS_ADMIN_KEY or ANIS_SHARE_ATTACKERS here: both confirmed dead
      # via grep, zero references anywhere in internal/config.go or the ANIS
      # client. A prior version of this file carried ANIS_ADMIN_KEY and
      # falsely claimed ARGUS "fails closed at startup" without it.)
      ANIS_ENABLED: ${ANIS_ENABLED:-true}
      ANIS_URL: https://anis.weboria.eu
      ANIS_LICENSE_KEY: ""
      DNS_SECURITY_LISTEN_ADDR: ":53"
    ports:
      - "127.0.0.1:${API_HOST_PORT:-9080}:8080"
      # DNS Security Engine. Host-only by default — a bare forwarding
      # resolver reachable from the network is exactly the profile abused
      # for DNS amplification attacks against third parties. Set
      # DNS_LISTEN_HOST to an internal interface IP to route real client
      # DNS traffic through it (never 0.0.0.0 on an untrusted network).
      - "${DNS_LISTEN_HOST:-127.0.0.1}:53:53/udp"
      - "${DNS_LISTEN_HOST:-127.0.0.1}:53:53/tcp"
    volumes:
      - nginx_data:/etc/nginx:rw
      - api_data:/app/data:rw
    depends_on:
      db:
        condition: service_healthy
      valkey:
        condition: service_started
      docker-socket-proxy:
        condition: service_started
    healthcheck:
      test: ["CMD", "wget", "-q", "--spider", "http://localhost:8080/health"]
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 360s

  # api-watchdog (2026-09-12, see docker/watchdog/README.md in the
  # argus-appliance repo): closes a real production gap -- `restart:
  # unless-stopped` above only ever reacts to the api container actually
  # EXITING. A real incident (CT 101, 2026-09-11/12) had it hang/wedge
  # internally for ~11 hours while `docker ps` still reported it `Up`; the
  # HEALTHCHECK above correctly kept detecting the hang the entire time,
  # but nothing was watching that status and acting on it. This polls the
  # same already-scoped docker-socket-proxy (zero new grants) and restarts
  # argus-api after several consecutive unhealthy checks.
  #
  # Built locally rather than pulled/pinned/cosign-verified like api/ui/
  # nginx above: there is not yet a released, signed argus-api-watchdog
  # image in the registry (tracked as a follow-up to extend ADR-0051's
  # verification set to a 4th component). The build context ships in this
  # repo so it stays fully reproducible from source, not a hand-built
  # local image nobody can regenerate.
  api-watchdog:
    build:
      context: ./watchdog
    container_name: argus-api-watchdog
    restart: unless-stopped
    security_opt:
      - "no-new-privileges:true"
    cap_drop:
      - ALL
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
    environment:
      WATCHDOG_TARGET_CONTAINER: argus-api
      WATCHDOG_INTERVAL_SECONDS: "30"
      WATCHDOG_UNHEALTHY_THRESHOLD: "5"
    depends_on:
      docker-socket-proxy:
        condition: service_started
      api:
        condition: service_started

  # React UI (Admin Panel), served over HTTPS
  ui:
    # See the api service's own comment above -- same ADR-0051 pinning.
    image: git-cloud.weboria.eu/weboria/argus-ui:${ARGUS_VERSION:-latest}
    container_name: argus-ui
    restart: unless-stopped
    security_opt:
      - "no-new-privileges:true"
    cap_drop:
      - ALL
    cap_add:
      - NET_BIND_SERVICE
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
    environment:
      TZ: ${TZ:-UTC}
    ports:
      - "${UI_PORT:-81}:443"
    volumes:
      - ui_data:/app/ssl:rw
    depends_on:
      - api

  # Nginx reverse proxy — host network mode, so real client IPs are visible
  # directly without needing PROXY protocol.
  nginx:
    # See the api service's own comment above -- same ADR-0051 pinning.
    image: git-cloud.weboria.eu/weboria/argus-proxy:${ARGUS_VERSION:-latest}
    container_name: argus-proxy
    restart: always
    network_mode: host
    # Not cap_drop'd: runs nginx in host-network mode with a root master that
    # drops workers to the nginx user, needing CHOWN/SETUID/SETGID plus binds
    # on privileged 80/443.
    security_opt:
      - "no-new-privileges:true"
    logging:
      driver: "json-file"
      options:
        max-size: "100m"
        max-file: "5"
    environment:
      TZ: ${TZ:-UTC}
    ulimits:
      nofile:
        soft: 65535
        hard: 65535
    volumes:
      - nginx_data:/etc/nginx:rw
    depends_on:
      - api
    healthcheck:
      test: ["CMD", "curl", "-f", "http://127.0.0.1:${NGINX_HTTP_PORT:-80}/health"]
      interval: 15s
      timeout: 5s
      retries: 3
      start_period: 10s

volumes:
  postgres_data:
    name: argus_postgres_data
  valkey_data:
    name: argus_valkey_data
  nginx_data:
    name: argus_nginx_data
  api_data:
    name: argus_api_data
  ui_data:
    name: argus_ui_data

networks:
  default:
    name: argus-network
    external: true