v1.1.1: harden deployment for easier fielding

Make a fresh deploy (especially Docker) work with no manual steps and behave
well on real-world hosts:

- Auto-scan skips subnets larger than /20 (e.g. a docker bridge 172.17.0.0/16
  = 65k hosts) so AUTO discovery no longer stalls startup or hammers the
  network on any host with Docker. Manual scan shares the same MIN_SCAN_PREFIX.
- Docker: config.json is auto-seeded from config.json.example by a new
  docker-entrypoint.sh and persisted on a ./data directory mount — no more
  single-file bind-mount footgun, no manual cp before first run.
- Config path is overridable via ADSBIT_CONFIG (defaults to ./config.json for
  bare metal; /app/data/config.json in the container).
- New /health endpoint (status/version/receivers/flights) + Docker HEALTHCHECK.
- Clean SIGTERM/SIGINT shutdown (docker stop / systemctl stop) — no traceback.
- .gitignore/.dockerignore exclude the data/ config volume.
- README/CLAUDE.md updated; VERSION bumped to 1.1.1.

Verified: image builds, fresh `docker run` seeds config and reports healthy,
oversized subnets skipped in-container, /health responds, graceful shutdown.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.dockerignore

@@ -3,6 +3,7 @@ __pycache__
 *.pyc
 *.pyo
 config.json
+data/
 screenshots/
 .claude/
 CLAUDE.md

.gitignore

@@ -26,3 +26,6 @@ Thumbs.db
 # Local config — contains the admin password hash and session secret.
 # Fresh installs start from config.json.example (see CONFIG.md).
 config.json
+
+# Docker config volume (holds the live config.json with secrets)
+data/

CLAUDE.md

@@ -18,6 +18,14 @@ python3 server.py
 
 Access at http://localhost:{web_port} (configured in config.json)
 
+### Deployment / fielding notes
+
+- **Config path** is overridable via the `ADSBIT_CONFIG` env var (defaults to `config.json` beside `server.py`). Docker uses `/app/data/config.json` on a mounted volume.
+- **`/health`** (no auth) returns status/version/receiver/flight counts — used by the Docker `HEALTHCHECK` and any external monitoring.
+- **`VERSION`** constant in `server.py` is the single source of truth; surfaced at startup, via `GET /api/config`, and in the admin header. Bump it for releases.
+- **Auto-scan** (`scan_for_receivers`) skips subnets larger than `MIN_SCAN_PREFIX` (/20) so a docker `/16` doesn't stall startup; the manual scan endpoint enforces the same limit.
+- **`docker-entrypoint.sh`** seeds `config.json` from `config.json.example` on first run, so a fresh `docker compose up` needs no manual steps. SIGTERM/SIGINT shut down cleanly (no traceback).
+
 ## Architecture
 
 ### Components

Dockerfile

@@ -13,9 +13,16 @@ RUN pip install --no-cache-dir -r requirements.txt && \
     rm -rf /var/lib/apt/lists/*
 
 COPY . .
+RUN chmod +x /app/docker-entrypoint.sh
 
 EXPOSE 2001
 
-ENV PYTHONUNBUFFERED=1
+# Persist config on a mounted volume by default (see docker-compose.yml).
+ENV PYTHONUNBUFFERED=1 \
+    ADSBIT_CONFIG=/app/data/config.json
 
-CMD ["python3", "server.py"]
+# Liveness probe against the unauthenticated /health endpoint.
+HEALTHCHECK --interval=30s --timeout=5s --start-period=20s --retries=3 \
+    CMD python3 -c "import urllib.request,sys; sys.exit(0 if urllib.request.urlopen('http://127.0.0.1:2001/health',timeout=4).status==200 else 1)" || exit 1
+
+ENTRYPOINT ["/app/docker-entrypoint.sh"]

README.md

@@ -44,14 +44,16 @@ On first run, visit http://localhost:2001 and the setup wizard will guide you th
 ### First Run
 
 ```bash
-# Create a config file (the setup wizard runs if you skip this)
+# Build and start — no config step needed
-cp config.json.example config.json
-
-# Build and start
 docker compose up -d
 ```
 
-The web UI is available at http://localhost:2001.
+A config file is created automatically in `data/` on first run, then the
+browser **setup wizard** walks you through receiver, location, and password
+setup. The web UI is available at http://localhost:2001.
+
+A `/health` endpoint (used by the container healthcheck) reports status,
+version, and receiver/flight counts: `curl http://localhost:2001/health`.
 
 ### Useful Commands
 
@@ -82,11 +84,11 @@ The following paths are bind-mounted from the repo directory and persist across
 
 | Path | Contents |
 |------|----------|
-| `config.json` | Server configuration and credentials |
+| `data/` | Server configuration and credentials (`config.json`, auto-seeded) |
 | `images/` | Aircraft and UI sprites (including uploads) |
 | `backgrounds/` | Theme background images (including custom themes) |
 
-Host networking (`network_mode: host`) is used so the server can auto-scan your LAN for ADS-B receivers.
+Host networking (`network_mode: host`) is used so the server can auto-scan your LAN for ADS-B receivers. Auto-scan skips oversized subnets (larger than /20, e.g. a Docker bridge `172.17.0.0/16`); if your receiver lives on such a network, set it explicitly via the admin Receivers tab.
 
 ## Requirements
 

docker-compose.yml

@@ -6,9 +6,11 @@ services:
     restart: unless-stopped
     environment:
       - PYTHONUNBUFFERED=1
+      - ADSBIT_CONFIG=/app/data/config.json
     volumes:
-      # Persist config (copy config.json.example to config.json before first run)
+      # Config persists here; auto-seeded from config.json.example on first run
-      - ./config.json:/app/config.json
+      # (a directory mount, so a fresh `docker compose up` just works).
+      - ./data:/app/data
       # Persist custom sprite uploads
       - ./images:/app/images
       # Persist custom theme backgrounds

docker-entrypoint.sh

@@ -0,0 +1,16 @@
+#!/bin/sh
+# Seed a config file on first run so `docker compose up` works on a fresh
+# clone with no manual steps. The config lives on a mounted data volume so it
+# persists across container rebuilds.
+set -e
+
+: "${ADSBIT_CONFIG:=/app/config.json}"
+CONFIG_DIR=$(dirname "$ADSBIT_CONFIG")
+mkdir -p "$CONFIG_DIR"
+
+if [ ! -f "$ADSBIT_CONFIG" ]; then
+    echo "[entrypoint] No config at $ADSBIT_CONFIG — seeding from config.json.example"
+    cp /app/config.json.example "$ADSBIT_CONFIG"
+fi
+
+exec python3 server.py

server.py

@@ -17,10 +17,16 @@ from aiohttp import web, ClientSession, ClientTimeout
 import netifaces
 import bcrypt
 
-VERSION = "1.1.0"
+VERSION = "1.1.1"
 
 WEB_DIR = Path(__file__).parent
-CONFIG_FILE = WEB_DIR / "config.json"
+# Config path is overridable (handy for containers persisting config on a
+# named volume); defaults to config.json next to this script for bare-metal.
+CONFIG_FILE = Path(os.environ.get("ADSBIT_CONFIG") or (WEB_DIR / "config.json"))
+
+# Smallest subnet prefix (largest network) we'll auto-scan: /20 = 4094 hosts.
+# Anything bigger (e.g. a docker /16) is skipped to keep scans fast and safe.
+MIN_SCAN_PREFIX = 20
 
 # Track server start time
 SERVER_START_TIME = time.time()
@@ -414,6 +420,15 @@ async def scan_for_receivers():
                 if ip and netmask and not ip.startswith('127.'):
                     try:
                         network = ipaddress.IPv4Network(f"{ip}/{netmask}", strict=False)
+                        # Skip subnets too large to scan quickly (e.g. a docker
+                        # bridge /16 = 65k hosts). Without this, AUTO discovery
+                        # hammers the network and stalls startup on any host with
+                        # Docker or a large/16. Use MANUAL or a custom subnet for
+                        # receivers that live on such networks.
+                        if network.prefixlen < MIN_SCAN_PREFIX:
+                            print(f"Skipping {network} on {iface}: too large "
+                                  f"({network.num_addresses} hosts) for auto-scan")
+                            continue
                         found.extend(await _scan_subnet(network, port))
                     except ValueError as e:
                         print(f"Invalid network {ip}/{netmask}: {e}")
@@ -984,9 +999,10 @@ async def handle_admin_scan_receivers(request):
             return web.json_response(
                 {"error": f"Invalid subnet: {subnet}"}, status=400)
 
-        if network.prefixlen < 20:
+        if network.prefixlen < MIN_SCAN_PREFIX:
             return web.json_response(
-                {"error": "Subnet too large (max /20 = 4096 hosts)"}, status=400)
+                {"error": f"Subnet too large (max /{MIN_SCAN_PREFIX} = "
+                          f"{2 ** (32 - MIN_SCAN_PREFIX) - 2} hosts)"}, status=400)
 
         port = config['receiver_port']
         found = await _scan_subnet(network, port)
@@ -1172,6 +1188,17 @@ async def handle_http(request):
     return web.Response(status=404, text="Not Found")
 
 
+async def handle_health(request):
+    """GET /health - liveness/readiness probe (no auth, no secrets)."""
+    return web.json_response({
+        "status": "ok",
+        "version": VERSION,
+        "receivers": len(receivers),
+        "flights": len(flights),
+        "uptime_seconds": int(time.time() - SERVER_START_TIME),
+    })
+
+
 # ---------------------------------------------------------------------------
 # Application setup
 # ---------------------------------------------------------------------------
@@ -1184,6 +1211,7 @@ async def start_http_server():
     app.router.add_get('/ws', websocket_handler)
 
     # Public API
+    app.router.add_get('/health', handle_health)
     app.router.add_get('/api/receiver-location', handle_receiver_location)
     app.router.add_get('/api/config', handle_config)
 
@@ -1282,4 +1310,11 @@ async def main():
 
 
 if __name__ == "__main__":
-    asyncio.run(main())
+    # Make SIGTERM (docker stop, systemctl stop) raise KeyboardInterrupt like
+    # SIGINT does, so shutdown is clean and quiet instead of a stack trace.
+    import signal
+    signal.signal(signal.SIGTERM, signal.default_int_handler)
+    try:
+        asyncio.run(main())
+    except (KeyboardInterrupt, SystemExit):
+        print("ADS-Bit server shutting down.")