Cluster Intranet Deployment Guide (Linux Docker)

This guide covers deploying LarkXR in a **Linux Docker cluster** configuration on an intranet. A minimum of two servers is required: one **master node** (management server, non-GPU) and one **GPU rendering node**.

## Prerequisites

1. **Environment Preparation** - Set up at least two servers, with at least one being a GPU server. Refer to [Environment Preparation](https://docs.paraverse.cc/doc/215/) for setup details.
   > If the master node will also serve as a rendering node, follow the [Stand-Alone Version deployment guide](https://docs.paraverse.cc/doc/243/) from Step 4 onward.
2. **Download Package** - Based on [System Components](https://docs.paraverse.cc/doc/239/), download and extract the Linux Docker package. Extract to a drive root directory with no Chinese characters in the path.
3. **Port Availability** - Ensure all required ports are free on both servers before proceeding.

## Required Ports

### Master Node

| Protocol | Port | Purpose |
|---|---|---|
| TCP | 8181 | Frontend page (nginx) and backend interface |
| TCP | 8182 | Backend page nginx forwarding |
| TCP | 8184 | Management service |
| TCP | 6379 | Redis cache |

### Rendering Node

| Protocol | Port | Purpose |
|---|---|---|
| TCP | 8089 | Sync and proxy service |
| TCP | 10002 | WebSocket communication |
| TCP | 10003 | Internal rendering server communication |
| TCP | 10004 | Rendering server SSL |
| UDP | 10011-10040 | Audio and video communication |

## Deployment Steps

### Master Node Setup

1. **Prepare the Linux server** - Set the open file limit and configure a static IP:
   ```
   ulimit -n 65535
   ```
   Set the system time to UTC.
2. **Upload the package** - Download `larkxr-center-ubuntu-docker` from the [Paraverse Dev Center](https://paraverse.cc/larkxr-resource/) and upload it to the server.
3. **Start services** - Navigate to the extracted directory and run:
   ```
   cd /opt/larkxr-center-ubuntu-docker-offline
   sh start.sh
   ```
4. **Verify the master node** - Open a browser and go to `http://<master-IP>:8182`. Log in with the default credentials:
   - **Username:** `admin`
   - **Password:** `123456`

### Rendering Node Setup

5. **Prepare the GPU server** - Ensure NVIDIA drivers are installed and verify with:
   ```
   nvidia-smi
   ```
   Set the system time to UTC and configure a static IP.
6. **(Optional) Kernel configuration** - To allow process tracing, edit `/etc/sysctl.d/10-ptrace.conf` and set:
   ```
   kernel.yama.ptrace_scope = 0
   ```
   Then apply the change:
   ```
   sudo /sbin/sysctl -p
   ```
7. **Upload the package** - Upload `larkxr-render-docker` to the rendering server.
8. **Start services** - Navigate to the extracted directory and run:
   ```
   cd /opt/larkxr-render-docker-offline
   sh start.sh
   ```
   To stop services, run `sh stop.sh`.

### Verification

9. **Confirm registration** - Log in to the master node backend at `http://<master-IP>:8182` and check the resource monitoring panel. A successfully connected GPU node should appear with a count of 1.