fix(rdesign): run Scribus runner as standalone supervisor process

The Scribus --python-script flag requires GUI initialization which
blocks in headless environments. Instead, run the runner as a separate
supervisor-managed Python process (always-on socket server). The bridge
server now simply verifies the socket exists rather than launching
Scribus.

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

docker/scribus-novnc/bridge/server.py

@@ -14,7 +14,6 @@ server translates HTTP requests into socket commands.
 import json
 import os
 import socket
-import subprocess
 import time
 from pathlib import Path
 from flask import Flask, request, jsonify
@@ -23,11 +22,7 @@ app = Flask(__name__)
 
 BRIDGE_SECRET = os.environ.get("BRIDGE_SECRET", "")
 SOCKET_PATH = "/tmp/scribus_bridge.sock"
-SCRIBUS_RUNNER = "/opt/bridge/scribus_runner.py"
 DESIGNS_DIR = Path("/data/designs")
-DISPLAY = os.environ.get("DISPLAY", ":1")
-
-_scribus_proc = None
 
 
 def _check_auth():
@@ -81,50 +76,24 @@ def before_request():
 @app.route("/health", methods=["GET"])
 def health():
     runner_alive = os.path.exists(SOCKET_PATH)
-    scribus_alive = _scribus_proc is not None and _scribus_proc.poll() is None
     return jsonify({
         "ok": True,
         "service": "scribus-bridge",
         "runner_connected": runner_alive,
-        "scribus_alive": scribus_alive,
     })
 
 
 @app.route("/api/scribus/start", methods=["POST"])
 def start_scribus():
-    """Launch Scribus with the bridge runner script."""
-    global _scribus_proc
-
-    if _scribus_proc and _scribus_proc.poll() is None:
+    """Verify runner socket is available. Optionally launch Scribus GUI for real rendering."""
+    # The runner process is managed by supervisor and should already be listening.
+    # Wait briefly for socket if it's still starting up.
+    for _ in range(10):
         if os.path.exists(SOCKET_PATH):
-            return jsonify({"ok": True, "message": "Scribus already running", "pid": _scribus_proc.pid})
-        # Process alive but runner socket missing — kill and restart
-        _scribus_proc.kill()
-        _scribus_proc.wait(timeout=5)
-        _scribus_proc = None
-
-    # Clean up stale socket
-    if os.path.exists(SOCKET_PATH):
-        os.remove(SOCKET_PATH)
-
-    env = os.environ.copy()
-    env["DISPLAY"] = DISPLAY
-
-    env["XDG_RUNTIME_DIR"] = "/tmp/runtime-root"
-    _scribus_proc = subprocess.Popen(
-        ["scribus", "--python-script", SCRIBUS_RUNNER],
-        env=env,
-        stdout=subprocess.PIPE,
-        stderr=subprocess.PIPE,
-    )
-
-    # Wait for runner socket to appear (up to 15s)
-    for _ in range(30):
-        if os.path.exists(SOCKET_PATH):
-            return jsonify({"ok": True, "message": "Scribus started", "pid": _scribus_proc.pid})
+            return jsonify({"ok": True, "message": "Runner connected", "runner_connected": True})
         time.sleep(0.5)
 
-    return jsonify({"ok": False, "error": "Scribus runner did not connect in time"}), 500
+    return jsonify({"ok": False, "error": "Runner socket not available. Check supervisor logs."}), 500
 
 
 @app.route("/api/scribus/command", methods=["POST"])

docker/scribus-novnc/supervisord.conf

@@ -20,6 +20,15 @@ autorestart=true
 priority=30
 startsecs=5
 
+[program:runner]
+command=python3 /opt/bridge/scribus_runner.py
+autorestart=true
+priority=35
+environment=DISPLAY=":1"
+stdout_logfile=/var/log/supervisor/runner.log
+stderr_logfile=/var/log/supervisor/runner_err.log
+startsecs=2
+
 [program:bridge]
 command=python3 /opt/bridge/server.py
 autorestart=true

modules/rdesign/design-agent-route.ts

@@ -61,7 +61,7 @@ async function bridgeState(): Promise<any> {
 	}
 }
 
-/** Start Scribus if not running, verify runner is connected. */
+/** Verify the bridge runner is connected and ready. */
 async function ensureScribusRunning(): Promise<any> {
 	const headers: Record<string, string> = { "Content-Type": "application/json" };
 	if (BRIDGE_SECRET) headers["X-Bridge-Secret"] = BRIDGE_SECRET;
@@ -69,28 +69,10 @@ async function ensureScribusRunning(): Promise<any> {
 	try {
 		const res = await fetch(`${SCRIBUS_BRIDGE_URL}/api/scribus/start`, {
 			method: "POST",
-			headers,
-			signal: AbortSignal.timeout(20_000),
-		});
-		const result = await res.json();
-		if (result.error) return result;
-
-		// Verify runner is actually connected by checking state
-		const stateRes = await fetch(`${SCRIBUS_BRIDGE_URL}/api/scribus/state`, {
 			headers,
 			signal: AbortSignal.timeout(10_000),
 		});
-		const state = await stateRes.json();
-		if (!state.error) return result;
-
-		// Runner not connected — force restart by calling start again
-		// (server.py now kills zombie Scribus when socket is missing)
-		const retryRes = await fetch(`${SCRIBUS_BRIDGE_URL}/api/scribus/start`, {
-			method: "POST",
-			headers,
-			signal: AbortSignal.timeout(20_000),
-		});
-		return await retryRes.json();
+		return await res.json();
 	} catch (e: any) {
 		return { error: `Bridge unreachable: ${e.message}` };
 	}