Docker 配置 OpenClaw

基础配置

首先下载官方仓库：

git clone https://github.com/openclaw/openclaw.git

然后设置 .env：

OPENCLAW_CONFIG_DIR=/mnt/8T/lc/open_claw_home/.openclaw
OPENCLAW_WORKSPACE_DIR=/mnt/8T/lc/open_claw_home/.openclaw/workspace
OPENCLAW_GATEWAY_PORT=18989
OPENCLAW_BRIDGE_PORT=18990
OPENCLAW_GATEWAY_BIND=lan
OPENCLAW_GATEWAY_TOKEN=efed7cf6abe0d8f2d4c9c59a35f4596540d9ac6b8ea6b76705079bc648dc0a87
OPENCLAW_IMAGE=openclaw:local
OPENCLAW_EXTRA_MOUNTS=/mnt/8T/lc/open_claw_home/playground:/home/node/playground:rw
OPENCLAW_HOME_VOLUME=
OPENCLAW_DOCKER_APT_PACKAGES="python3 python3-pip wget curl git"
OPENCLAW_EXTENSIONS=

# 开启沙箱功能
OPENCLAW_SANDBOX=1
OPENCLAW_DOCKER_SOCKET=/var/run/docker.sock
DOCKER_GID=
OPENCLAW_INSTALL_DOCKER_CLI=1
OPENCLAW_ALLOW_INSECURE_PRIVATE_WS=

修改 docker-setup.sh 的开头部分：

#!/usr/bin/env bash
set -euo pipefail
# 1. 定义根目录
ROOT_DIR="/mnt/8T/lc/open_claw/openclaw/"

# 2. 强力加载 .env (支持带空格的变量)
if [ -f "$ROOT_DIR/.env" ]; then
    echo "==> Loading configuration from $ROOT_DIR/.env"
    # 使用 allexport 模式安全读取
    set -a
    source "$ROOT_DIR/.env"
    set +a
else
    echo "WARNING: .env file not found at $ROOT_DIR/.env"
fi

然后配置Dokcker 镜像：

bash <(wget -qO- https://xuanyuan.cloud/docker.sh)

运行 docker-setup.sh。然后这个时候会一直重启，就进入前文映射的/open_claw_home/.openclaw修改配置文件：

"gateway": {
  "controlUi": {
    "dangerouslyAllowHostHeaderOriginFallback": true
  }
}z

然后阿里云会给你新的配置文件，千万不要直接复制粘贴。只把 models 和 agents 两个字段复制粘贴即可。

阿里云配置步骤：https://help.aliyun.com/zh/model-studio/openclaw-coding-plan

如何配置 GPU 环境？

安装环境：

sudo apt update
sudo apt install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

然后查看：

sudo docker info | grep -i runtime

看到 nvidia 字样即可。

测试 GPU 容器：

docker run --rm --gpus all nvidia/cuda:12.2.0-base-ubuntu22.04 nvidia-smi

然后修改 docker-compose.yml：

services:
  openclaw-gateway:
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]