[WebRTC_DEMO] 部署成功后访问页面后无法开启start call

[xiaojiwang]

三方gpu服务器环境（基于容器的）：
Ubuntu 22.04
Cuda 13.0
python 3.12

利用oneclick.sh部署成功，防火墙开通了8088、7880端口，浏览器能够访问demo页面，但是点击start call等待一段时间后报错：

启动日志如下：
`

MiniCPM-o WebRTC Demo One-Click Start (without Docker)

[INFO] Running preflight checks...
[OK] livekit-server livekit-server version 1.9.11
[OK] Python: Python 3.12.11 (/usr/local/miniconda3/envs/py312/bin/python)
[OK] Node.js: v20.5.0
[OK] pnpm: 10.30.2
[OK] ffmpeg: ffmpeg version 4.4.2-0ubuntu0.22.04.1 Copyright (c) 2000-2021 the FFmpeg developers
[OK] llama-server: /home/soft/llama.cpp-omni/build/bin/llama-server
[OK] Model directory: /home/soft/models/openbmb/MiniCPM-o-4_5-gguf
[OK] Preflight checks passed

[INFO] ========== [1/4] Starting LiveKit Server ==========
[INFO] Detected local IP: 10.60.41.116
[OK] Updated livekit.yaml node_ip and domain to 10.60.41.116
[OK] LiveKit is running (port 7880, waited 1s)

[INFO] ========== [2/4] Starting Backend (FastAPI) ==========
[OK] Backend Python dependencies are ready
[OK] Backend is running (port 8021, waited 5s)
[OK] Backend health check passed: {"status":"healthy","service":"minicpmo-backend"}

[INFO] ========== [3/4] Starting C++ Inference Service (model loading takes 2-3 min) ==========
[INFO] Mode: duplex | Vision backend: default | Port: 9060
[OK] C++ Inference (health check port) is running (port 9061, waited 1s)
[INFO] Waiting for C++ Inference Service health check to pass (max 300s)...
[INFO] Still waiting for C++ Inference Service to start... (30s/300s)
[OK] C++ Inference Service health check passed (waited 55s)
[OK] C++ inference service registered with Backend (status: available)

[INFO] ========== [4/4] Starting Frontend (Vue + Vite) ==========
[OK] Frontend dependencies are ready (node_modules exists)
[OK] HTTPS certificate is ready
[INFO] Frontend will reflect CPP_MODE=duplex (unavailable mode tab shown as disabled)
[INFO] Frontend mode: production build (prod)
[OK] Frontend already built (dist/ exists, set FORCE_BUILD=1 to force rebuild)
[OK] Frontend is running (port 8088, waited 1s)

==============================================
[OK] All services started successfully!

[INFO] Service status:

● livekit running PID=211 port=7880
● backend running PID=239 port=8021
● cpp_server running PID=321 port=9060
● frontend running PID=455 port=8088

[INFO] Log files:
LiveKit: /home/soft/.logs/livekit.log
Backend: /home/soft/.logs/backend.log
C++ Inference: /home/soft/.logs/cpp_server.log
Frontend: /home/soft/.logs/frontend.log

[INFO] Access URLs: (frontend mode: prod)
Frontend: https://10.60.41.116:8088 (accept self-signed certificate on first visit)
Backend API: http://10.60.41.116:8021
LiveKit: ws://10.60.41.116:7880
Inference: http://10.60.41.116:9060
Inference Health: http://10.60.41.116:9061/health
`