背景

Win11的docker desktop安装open-notebook报错看不懂；Win11用源码安装报缺少SurrealDB的配置。

故用WSL2 + WSL的Docker安装。

WSL的docker安装open-notebook

官方安装原文：Single-Container Deployment (Recommended for Beginners)

单容器部署（推荐初学者使用）

非常适合个人使用或 PikaPods 等平台：

创建项目目录

mkdir open-notebook
cd open-notebook

创建 Docker Compose 文件

# 创建 docker-compose.yml
cat > docker-compose.yml << 'EOF'
services:
  open_notebook:
    image: lfnovo/open_notebook:v1-latest-single
    ports:
      - "8502:8502"
      - "5055:5055"
    env_file:
      - ./docker.env
    pull_policy: always
    volumes:
      - ./notebook_data:/app/data
      - ./surreal_single_data:/mydata
    restart: always
EOF

创建环境配置文件

# 创建 docker.env 文件并填入您的 API 密钥
cat > docker.env << 'EOF'
# 必需：至少配置一个 AI 服务提供商
# OPENAI_API_KEY=您的-openai-密钥-在此填写
# 或者

# 数据库设置（请勿修改）
SURREAL_ADDRESS=localhost
SURREAL_PORT=8000
SURREAL_USER=root
SURREAL_PASS=root
SURREAL_NAMESPACE=open_notebook
SURREAL_DATABASE=production

# 可选：密码保护功能
# OPEN_NOTEBOOK_PASSWORD=您的安全密码
EOF

参考.env文件（官方配置文件.env.example示例），可以从里面获取自己的API供应商配置格式，比如OPENAI_COMPATIBLEOpen API兼容格式的：

# API CONFIGURATION
# URL where the API can be accessed by the browser
# This setting allows the frontend to connect to the API at runtime (no rebuild needed!)
#
# IMPORTANT: Do NOT include /api at the end - it will be added automatically!
#
# Common scenarios:
# - Docker on localhost: http://localhost:5055 (default, works for most cases)
# - Docker on LAN/remote server: http://192.168.1.100:5055 or http://your-server-ip:5055
# - Behind reverse proxy with custom domain: https://your-domain.com
# - Behind reverse proxy with subdomain: https://api.your-domain.com
#
# Examples for reverse proxy users:
# - API_URL=https://notebook.example.com  (frontend will call https://notebook.example.com/api/*)
# - API_URL=https://api.example.com       (frontend will call https://api.example.com/api/*)
#
# Note: If not set, the system will auto-detect based on the incoming request.
# Only set this if you need to override the auto-detection (e.g., reverse proxy scenarios).
API_URL=http://localhost:5055

# INTERNAL API URL (Server-Side)
# URL where Next.js server-side should proxy API requests (via rewrites)
# This is DIFFERENT from API_URL which is used by the browser client
#
# INTERNAL_API_URL is used by Next.js rewrites to forward /api/* requests to the FastAPI backend
# API_URL is used by the browser to know where to make API calls
#
# Default: http://localhost:5055 (single-container deployment - both services on same host)
# Override for multi-container: INTERNAL_API_URL=http://api-service:5055
#
# Common scenarios:
# - Single container (default): Don't set - defaults to http://localhost:5055
# - Multi-container Docker Compose: INTERNAL_API_URL=http://api:5055 (use service name)
# - Kubernetes/advanced networking: INTERNAL_API_URL=http://api-service.namespace.svc.cluster.local:5055
#
# Why two variables?
# - API_URL: External/public URL that browsers use (can be https://your-domain.com)
# - INTERNAL_API_URL: Internal container networking URL (usually http://localhost:5055 or service name)
#
# INTERNAL_API_URL=http://localhost:5055

# API CLIENT TIMEOUT (in seconds)
# Controls how long the frontend/Streamlit UI waits for API responses
# Increase this if you're using slow AI providers or hardware (Ollama on CPU, remote LM Studio, etc.)
# Default: 300 seconds (5 minutes) - sufficient for most transformation/insight operations
#
# Common scenarios:
# - Fast cloud APIs (OpenAI, Anthropic): 300 seconds is more than enough
# - Local Ollama on GPU: 300 seconds should work fine
# - Local Ollama on CPU: Consider 600 seconds (10 minutes) or more
# - Remote LM Studio over slow network: Consider 900 seconds (15 minutes)
# - Very large documents: May need 900+ seconds
#
# API_CLIENT_TIMEOUT=300

# ESPERANTO LLM TIMEOUT (in seconds)
# Controls the timeout for AI model API calls at the Esperanto library level
# This is separate from API_CLIENT_TIMEOUT and applies to the actual LLM provider requests
# Only increase this if you're experiencing timeouts during model inference itself
# Default: 60 seconds (built into Esperanto)
#
# Important: This should generally be LOWER than API_CLIENT_TIMEOUT to allow proper error handling
#
# Common scenarios:
# - Fast cloud APIs (OpenAI, Anthropic, Groq): 60 seconds is sufficient
# - Local Ollama with small models: 120-180 seconds may help
# - Local Ollama with large models on CPU: 300+ seconds
# - Remote or self-hosted LLMs: 180-300 seconds depending on hardware
#
# Note: If transformations complete but you see timeout errors, increase API_CLIENT_TIMEOUT first.
# Only increase ESPERANTO_LLM_TIMEOUT if the model itself is timing out during inference.
#
# ESPERANTO_LLM_TIMEOUT=60

# SECURITY
# Set this to protect your Open Notebook instance with a password (for public hosting)
# OPEN_NOTEBOOK_PASSWORD=

# OPENAI
# OPENAI_API_KEY=


# ANTHROPIC
# ANTHROPIC_API_KEY=

# GEMINI
# this is the best model for long context and podcast generation
# GOOGLE_API_KEY=
# GEMINI_API_BASE_URL=  # Optional: Override default endpoint (for Vertex AI, proxies, etc.)

# VERTEXAI
# VERTEX_PROJECT=my-google-cloud-project-name
# GOOGLE_APPLICATION_CREDENTIALS=./google-credentials.json
# VERTEX_LOCATION=us-east5

# MISTRAL
# MISTRAL_API_KEY=

# DEEPSEEK
# DEEPSEEK_API_KEY=

# OLLAMA
# OLLAMA_API_BASE="http://10.20.30.20:11434"

# OPEN ROUTER
# OPENROUTER_BASE_URL="https://openrouter.ai/api/v1"
# OPENROUTER_API_KEY=

# GROQ
# GROQ_API_KEY=

# XAI
# XAI_API_KEY=

# ELEVENLABS
# Used only by the podcast feature
# ELEVENLABS_API_KEY=

# TTS BATCH SIZE
# Controls concurrent TTS requests for podcast generation (default: 5)
# Lower values reduce provider load but increase generation time
# Recommended: OpenAI=5, ElevenLabs=2, Google=4, Custom=1
# TTS_BATCH_SIZE=2

# VOYAGE AI
# VOYAGE_API_KEY=

# OPENAI COMPATIBLE ENDPOINTS
# Generic configuration (applies to all modalities: language, embedding, STT, TTS)
# OPENAI_COMPATIBLE_BASE_URL=
# OPENAI_COMPATIBLE_API_KEY=

# Mode-specific configuration (overrides generic if set)
# Use these when you want different endpoints for different capabilities
# OPENAI_COMPATIBLE_BASE_URL_LLM=http://localhost:1234/v1
# OPENAI_COMPATIBLE_API_KEY_LLM=
# OPENAI_COMPATIBLE_BASE_URL_EMBEDDING=http://localhost:8080/v1
# OPENAI_COMPATIBLE_API_KEY_EMBEDDING=
# OPENAI_COMPATIBLE_BASE_URL_STT=http://localhost:9000/v1
# OPENAI_COMPATIBLE_API_KEY_STT=
# OPENAI_COMPATIBLE_BASE_URL_TTS=http://localhost:9000/v1
# OPENAI_COMPATIBLE_API_KEY_TTS=

# AZURE OPENAI
# AZURE_OPENAI_API_KEY=
# AZURE_OPENAI_ENDPOINT=
# AZURE_OPENAI_API_VERSION="2024-12-01-preview"
# AZURE_OPENAI_DEPLOYMENT_NAME=

# USE THIS IF YOU WANT TO DEBUG THE APP ON LANGSMITH
# LANGCHAIN_TRACING_V2=true
# LANGCHAIN_ENDPOINT="https://api.smith.langchain.com"
# LANGCHAIN_API_KEY=
# LANGCHAIN_PROJECT="Open Notebook"

# CONNECTION DETAILS FOR YOUR SURREAL DB
# New format (preferred) - WebSocket URL
SURREAL_URL="ws://surrealdb/rpc:8000"
SURREAL_USER="root"
SURREAL_PASSWORD="root"
SURREAL_NAMESPACE="open_notebook"
SURREAL_DATABASE="staging"

# OPEN_NOTEBOOK_PASSWORD=

# FIRECRAWL - Get a key at https://firecrawl.dev/
FIRECRAWL_API_KEY=

# JINA - Get a key at https://jina.ai/
JINA_API_KEY=

启动 Open Notebook

docker compose up -d

问题解决

docker启动报错

问题：

  > ✘ open_notebook Error Get "https://registry-1.docker.io/v2/": EOF                                                 5.2s  
  Error response from daemon: Get "https://registry-1.docker.io/v2/": EOF  

把VPN开全局，因为Docker不对中国大陆提供服务

wsl报找不到docker

命令和报错信息：

docker compose up -d

The command 'docker' could not be found in this WSL 2 distro.
We recommend to activate the WSL integration in Docker Desktop settings.

For details about using Docker Desktop with WSL 2, visit:

https://docs.docker.com/go/wsl2/

需要在windows上先开Docker Desktop，然后再在wsl命令行里输入docker命令。