Fix SSL cert creation and HTTP fallback for Unauthenticated error

- Create NPM proxy host WITHOUT SSL initially (ssl_forced=False),
  then request Let's Encrypt cert, then enable SSL only after cert
  is assigned. Prevents broken proxy when cert fails.
- If SSL cert creation fails, automatically fall back to HTTP mode:
  re-render management.json, dashboard.env, relay.env with http://
  URLs and recreate containers so dashboard login works.
- Better error logging in _request_ssl with specific timeout hints.
- Use template variables for relay WebSocket protocol (rels/rel)
  instead of hardcoded rels:// in management.json.j2 and relay.env.j2.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app/services/netbird_service.py

@@ -119,10 +119,12 @@ async def deploy_customer(db: Session, customer_id: int) -> dict[str, Any]:
             external_url = f"http://localhost:{dashboard_port}"
             netbird_protocol = "http"
             netbird_port = str(dashboard_port)
+            relay_ws_protocol = "rel"
         else:
             external_url = f"https://{netbird_domain}"
             netbird_protocol = "https"
             netbird_port = "443"
+            relay_ws_protocol = "rels"
 
         # Step 4: Create instance directory
         instance_dir = os.path.join(config.data_dir, f"kunde{customer_id}")
@@ -151,6 +153,7 @@ async def deploy_customer(db: Session, customer_id: int) -> dict[str, Any]:
             "netbird_dashboard_image": config.netbird_dashboard_image,
             "docker_network": config.docker_network,
             "datastore_encryption_key": datastore_key,
+            "relay_ws_protocol": relay_ws_protocol,
         }
 
         _render_template(jinja_env, "docker-compose.yml.j2",
@@ -241,9 +244,11 @@ async def deploy_customer(db: Session, customer_id: int) -> dict[str, Any]:
                     f"NPM proxy creation failed: {npm_result['error']}",
                 )
             else:
+                ssl_ok = npm_result.get("ssl", False)
                 _log_action(
                     db, customer_id, "deploy", "info",
-                    f"NPM proxy host created: {netbird_domain} -> {forward_host}:{dashboard_port}",
+                    f"NPM proxy host created: {netbird_domain} -> {forward_host}:{dashboard_port} "
+                    f"(SSL: {'OK' if ssl_ok else 'FAILED — check DNS and port 80 accessibility'})",
                 )
 
             # Create NPM UDP stream for relay STUN port
@@ -267,6 +272,36 @@ async def deploy_customer(db: Session, customer_id: int) -> dict[str, Any]:
                     f"NPM UDP stream created: port {allocated_port} -> {forward_host}:{allocated_port}",
                 )
 
+            # Step 9b: If SSL failed, fall back to HTTP so the dashboard works
+            ssl_ok = npm_result.get("ssl", False) if not npm_result.get("error") else False
+            if not ssl_ok:
+                logger.warning("SSL cert failed for %s — switching configs to HTTP", netbird_domain)
+                external_url = f"http://{netbird_domain}"
+                netbird_protocol = "http"
+                netbird_port = "80"
+                relay_ws_protocol = "rel"
+                template_vars["external_url"] = external_url
+                template_vars["netbird_protocol"] = netbird_protocol
+                template_vars["netbird_port"] = netbird_port
+                template_vars["relay_ws_protocol"] = relay_ws_protocol
+
+                # Re-render configs that contain URL/protocol references
+                _render_template(jinja_env, "management.json.j2",
+                                 os.path.join(instance_dir, "management.json"), **template_vars)
+                _render_template(jinja_env, "dashboard.env.j2",
+                                 os.path.join(instance_dir, "dashboard.env"), **template_vars)
+                _render_template(jinja_env, "relay.env.j2",
+                                 os.path.join(instance_dir, "relay.env"), **template_vars)
+
+                # Recreate containers to pick up new config
+                docker_service.compose_up(instance_dir, container_prefix, timeout=120)
+
+                _log_action(
+                    db, customer_id, "deploy", "info",
+                    "SSL not available — switched to HTTP mode. "
+                    "To enable HTTPS: ensure DNS resolves and port 80 is reachable, then re-deploy.",
+                )
+
         # Step 10: Create deployment record
         setup_url = external_url
 

app/services/npm_service.py

@@ -127,6 +127,10 @@ async def create_proxy_host(
     Caddy reverse proxy is listening. Caddy handles internal routing to
     management, signal, relay, and dashboard containers.
 
+    Creates the proxy host WITHOUT SSL first (so HTTP works immediately),
+    then requests a Let's Encrypt certificate, and only enables SSL
+    after the cert is successfully assigned.
+
     Args:
         api_url: NPM API base URL.
         npm_email: NPM login email.
@@ -137,16 +141,18 @@ async def create_proxy_host(
         admin_email: Email for Let's Encrypt.
 
     Returns:
-        Dict with ``proxy_id`` on success or ``error`` on failure.
+        Dict with ``proxy_id`` and ``ssl`` (bool) on success, or ``error`` on failure.
     """
+    # Step 1: Create proxy host WITHOUT SSL — so HTTP works immediately
+    # SSL is enabled later only after a cert is successfully obtained.
     payload = {
         "domain_names": [domain],
         "forward_scheme": "http",
         "forward_host": forward_host,
         "forward_port": forward_port,
         "certificate_id": 0,
-        "ssl_forced": True,
-        "hsts_enabled": True,
+        "ssl_forced": False,
+        "hsts_enabled": False,
         "hsts_subdomains": False,
         "http2_support": True,
         "block_exploits": True,
@@ -162,14 +168,12 @@ async def create_proxy_host(
 
     try:
         async with httpx.AsyncClient(timeout=180) as client:  # Long timeout for LE cert
-            # Step 1: Login to NPM
             token = await _npm_login(client, api_url, npm_email, npm_password)
             headers = {
                 "Authorization": f"Bearer {token}",
                 "Content-Type": "application/json",
             }
 
-            # Step 2: Create proxy host
             resp = await client.post(
                 f"{api_url}/nginx/proxy-hosts", json=payload, headers=headers
             )
@@ -179,10 +183,10 @@ async def create_proxy_host(
                 logger.info("Created NPM proxy host %s -> %s:%d (id=%s)",
                             domain, forward_host, forward_port, proxy_id)
 
-                # Step 3: Request SSL certificate
-                await _request_ssl(client, api_url, headers, proxy_id, domain, admin_email)
+                # Step 2: Request SSL certificate and enable HTTPS
+                ssl_ok = await _request_ssl(client, api_url, headers, proxy_id, domain, admin_email)
 
-                return {"proxy_id": proxy_id}
+                return {"proxy_id": proxy_id, "ssl": ssl_ok}
             else:
                 error_msg = f"NPM returned {resp.status_code}: {resp.text[:300]}"
                 logger.error("Failed to create proxy host: %s", error_msg)
@@ -202,11 +206,13 @@ async def _request_ssl(
     proxy_id: int,
     domain: str,
     admin_email: str,
-) -> None:
-    """Request a Let's Encrypt SSL certificate for a proxy host.
+) -> bool:
+    """Request a Let's Encrypt SSL certificate and enable HTTPS on the proxy host.
 
-    Let's Encrypt validation can take up to 120 seconds, so we use
-    a longer timeout for certificate requests.
+    Flow:
+    1. Create LE certificate via NPM API (HTTP-01 validation, up to 120s)
+    2. Assign certificate to the proxy host
+    3. Enable ssl_forced + hsts on the proxy host
 
     Args:
         client: httpx client (already authenticated).
@@ -215,7 +221,14 @@ async def _request_ssl(
         proxy_id: The proxy host ID.
         domain: The domain to certify.
         admin_email: Contact email for LE.
+
+    Returns:
+        True if SSL was successfully enabled, False otherwise.
     """
+    if not admin_email:
+        logger.warning("No admin email set — skipping SSL certificate for %s", domain)
+        return False
+
     ssl_payload = {
         "domain_names": [domain],
         "provider": "letsencrypt",
@@ -227,30 +240,57 @@ async def _request_ssl(
         },
     }
     try:
-        logger.info("Requesting Let's Encrypt certificate for %s ...", domain)
+        logger.info("Requesting Let's Encrypt certificate for %s (email: %s) ...", domain, admin_email)
         resp = await client.post(
             f"{api_url}/nginx/certificates",
             json=ssl_payload,
             headers=headers,
             timeout=120,  # LE validation can be slow
         )
-        if resp.status_code in (200, 201):
+        if resp.status_code not in (200, 201):
+            logger.error(
+                "SSL cert request for %s failed (HTTP %s): %s",
+                domain, resp.status_code, resp.text[:500],
+            )
+            return False
+
         cert_id = resp.json().get("id")
-            logger.info("Certificate created (id=%s), assigning to proxy host %s", cert_id, proxy_id)
+        logger.info("Certificate created (id=%s) for %s", cert_id, domain)
+
+        # Assign cert AND enable SSL + HSTS in one update
+        ssl_update = {
+            "certificate_id": cert_id,
+            "ssl_forced": True,
+            "hsts_enabled": True,
+            "http2_support": True,
+        }
         assign_resp = await client.put(
             f"{api_url}/nginx/proxy-hosts/{proxy_id}",
-                json={"certificate_id": cert_id},
+            json=ssl_update,
             headers=headers,
         )
         if assign_resp.status_code in (200, 201):
-                logger.info("SSL certificate %s assigned to proxy host %s", cert_id, proxy_id)
+            logger.info("SSL enabled on proxy host %s for %s (cert_id=%s)", proxy_id, domain, cert_id)
+            return True
         else:
-                logger.warning("Failed to assign cert to proxy host: %s %s",
-                               assign_resp.status_code, assign_resp.text[:200])
-        else:
-            logger.warning("SSL cert request returned %s: %s", resp.status_code, resp.text[:500])
+            logger.error(
+                "Failed to assign cert %s to proxy host %s: HTTP %s — %s",
+                cert_id, proxy_id, assign_resp.status_code, assign_resp.text[:300],
+            )
+            return False
+
+    except httpx.TimeoutException:
+        logger.error(
+            "SSL cert request for %s timed out after 120s. "
+            "Check: 1) DNS resolves %s to your server, "
+            "2) Port 80 is accessible from the internet, "
+            "3) NPM is listening on port 80.",
+            domain, domain,
+        )
+        return False
     except Exception as exc:
-        logger.warning("SSL certificate request failed: %s", exc)
+        logger.error("SSL certificate request failed for %s: %s", domain, exc)
+        return False
 
 
 async def create_stream(

templates/management.json.j2

@@ -22,7 +22,7 @@
   },
   "Relay": {
     "Addresses": [
-      "rels://{{ netbird_domain }}:443"
+      "{{ relay_ws_protocol }}://{{ netbird_domain }}:{{ netbird_port }}"
     ],
     "CredentialsTTL": "24h",
     "Secret": "{{ relay_secret }}"

templates/relay.env.j2

@@ -2,6 +2,6 @@
 # {{ subdomain }}.{{ base_domain }}
 NB_AUTH_SECRET={{ relay_secret }}
 NB_LISTEN_ADDRESS=:80
-NB_EXPOSED_ADDRESS=rels://{{ subdomain }}.{{ base_domain }}:443
+NB_EXPOSED_ADDRESS={{ relay_ws_protocol }}://{{ subdomain }}.{{ base_domain }}:{{ netbird_port }}
 NB_ENABLE_STUN=true
 NB_STUN_PORTS=3478