新增内容: - 完整的架构文档 (docs/MEM0_ARCHITECTURE.md) - 部署指南 (docs/MEM0_DEPLOYMENT.md) - 部署脚本 (scripts/01-12.sh) - mem0 Integration Skill (skills/mem0-integration/) - 备份和监控脚本 已完成功能: ✅ Tailscale 组网 (100.115.94.1) ✅ Qdrant Master 向量数据库 ✅ Dozzle 日志系统 ✅ mem0 记忆管理 ✅ OpenClaw /memory 命令 ✅ 自动备份系统 ✅ 监控系统 架构特点: - 中心化部署，支持多 Agent 扩展 - Tailscale 内网安全通信 - 分层记忆架构 (本地 + 共享) - 一键部署和回滚 - 完整的文档和脚本master
Eason (陈医生)
1 month ago
parent
3ad7de00b8
commit
7036390772
19 changed files with 1503 additions and 0 deletions
Binary file not shown.
@ -0,0 +1,427 @@ |
||||
# mem0 记忆系统架构文档 |
||||
|
||||
## 版本信息 |
||||
- **文档版本**: 1.0.0 |
||||
- **创建日期**: 2026-02-22 |
||||
- **最后更新**: 2026-02-22 |
||||
- **部署环境**: Ubuntu 24.04 LTS, Docker 29.2.1 |
||||
|
||||
--- |
||||
|
||||
## 1. 系统架构 |
||||
|
||||
### 1.1 整体架构图 |
||||
|
||||
``` |
||||
┌─────────────────────────────────────────────────────────────────┐ |
||||
│ Tailscale 虚拟内网 │ |
||||
│ │ |
||||
│ ┌─────────────────────────────────────────────────────────┐ │ |
||||
│ │ 中心节点 (vps-vaym) │ │ |
||||
│ │ Tailscale IP: 100.115.94.1 │ │ |
||||
│ │ 节点名称：mem0-general-center │ │ |
||||
│ │ │ │ |
||||
│ │ ┌──────────────────────────────────────────────────┐ │ │ |
||||
│ │ │ Docker Compose Stack │ │ │ |
||||
│ │ │ ┌──────────────┐ ┌──────────────┐ │ │ │ |
||||
│ │ │ │ Qdrant │ │ Dozzle │ │ │ │ |
||||
│ │ │ │ Master │ │ (日志) │ │ │ │ |
||||
│ │ │ │ :6333 │ │ :9999 │ │ │ │ |
||||
│ │ │ └──────────────┘ └──────────────┘ │ │ │ |
||||
│ │ └──────────────────────────────────────────────────┘ │ │ |
||||
│ │ │ │ |
||||
│ │ ┌──────────────────────────────────────────────────┐ │ │ |
||||
│ │ │ OpenClaw + mem0 Integration │ │ │ |
||||
│ │ │ Gateway: 18789 │ │ │ |
||||
│ │ │ Skill: mem0-integration │ │ │ |
||||
│ │ └──────────────────────────────────────────────────┘ │ │ |
||||
│ └─────────────────────────────────────────────────────────┘ │ |
||||
│ │ |
||||
│ 未来扩展： │ |
||||
│ ┌──────────────┐ ┌──────────────┐ ┌──────────────┐ │ |
||||
│ │ Agent-1 │ │ Agent-2 │ │ Agent-N │ │ |
||||
│ │ (crypto) │ │ (advert) │ │ (life) │ │ |
||||
│ │ 100.64.x.x │ │ 100.64.x.x │ │ 100.64.x.x │ │ |
||||
│ └──────────────┘ └──────────────┘ └──────────────┘ │ |
||||
└─────────────────────────────────────────────────────────────────┘ |
||||
``` |
||||
|
||||
### 1.2 技术栈 |
||||
|
||||
| 组件 | 技术 | 版本 | 用途 | |
||||
|------|------|------|------| |
||||
| **组网** | Tailscale | 1.94.2 | 虚拟内网，安全通信 | |
||||
| **向量数据库** | Qdrant | 1.15.3 | 记忆存储和检索 | |
||||
| **记忆系统** | mem0 | 1.0.4 | 记忆管理层 | |
||||
| **LLM** | DashScope (Qwen) | - | 记忆处理和 embedding | |
||||
| **容器** | Docker | 29.2.1 | 服务隔离 | |
||||
| **日志** | Dozzle | latest | 实时日志查看 | |
||||
|
||||
### 1.3 数据流 |
||||
|
||||
``` |
||||
用户 → OpenClaw → mem0 Client → Qdrant Local → (异步同步) → Qdrant Master |
||||
↓ |
||||
DashScope API (Embedding + LLM) |
||||
``` |
||||
|
||||
--- |
||||
|
||||
## 2. 部署清单 |
||||
|
||||
### 2.1 中心节点 (vps-vaym) |
||||
|
||||
**已安装服务**： |
||||
- ✅ Tailscale (系统级) |
||||
- ✅ Qdrant Master (Docker) |
||||
- ✅ Dozzle (Docker) |
||||
- ✅ mem0 Integration (OpenClaw Skill) |
||||
|
||||
**目录结构**： |
||||
``` |
||||
/opt/mem0-center/ |
||||
├── docker-compose.yml # Docker 配置 |
||||
├── .env # 环境变量 |
||||
├── qdrant_storage/ # Qdrant 数据 |
||||
├── snapshots/ # Qdrant 快照 |
||||
├── tailscale/ # Tailscale 状态 |
||||
├── logs/ # 日志 |
||||
└── backup/ # 备份 |
||||
|
||||
/root/.openclaw/workspace/ |
||||
├── scripts/ # 部署脚本 |
||||
│ ├── 01-system-check.sh |
||||
│ ├── 02-install-tailscale.sh |
||||
│ ├── 03-create-directories.sh |
||||
│ ├── 05-start-center.sh |
||||
│ ├── 06-create-mem0-skill.sh |
||||
│ ├── 07-install-dependencies.sh |
||||
│ ├── 10-create-backup.sh |
||||
│ └── 12-monitoring.sh |
||||
├── skills/mem0-integration/ |
||||
│ ├── SKILL.md |
||||
│ ├── skill.json |
||||
│ ├── config.yaml |
||||
│ ├── mem0_client.py |
||||
│ ├── commands.py |
||||
│ └── openclaw_commands.py |
||||
├── docs/ |
||||
│ └── MEM0_DEPLOYMENT.md |
||||
└── backup/ # 备份文件 |
||||
``` |
||||
|
||||
### 2.2 网络配置 |
||||
|
||||
**Tailscale**： |
||||
- 网络名称：mem0-general-center |
||||
- 业务类型：general |
||||
- 节点角色：center |
||||
- IP 地址：100.115.94.1 |
||||
|
||||
**端口映射**： |
||||
| 服务 | 容器端口 | 主机端口 | 访问范围 | |
||||
|------|---------|---------|---------| |
||||
| Qdrant | 6333 | 127.0.0.1:6333 | 仅本地 + Tailscale | |
||||
| Dozzle | 8080 | 127.0.0.1:9999 | 仅本地 + Tailscale | |
||||
| OpenClaw | 18789 | 18789 | 本地 | |
||||
|
||||
--- |
||||
|
||||
## 3. 配置详情 |
||||
|
||||
### 3.1 Docker Compose 配置 |
||||
|
||||
文件：`/opt/mem0-center/docker-compose.yml` |
||||
|
||||
```yaml |
||||
version: '3.8' |
||||
services: |
||||
qdrant-master: |
||||
image: qdrant/qdrant:v1.15.3 |
||||
ports: |
||||
- "127.0.0.1:6333:6333" |
||||
volumes: |
||||
- ./qdrant_storage:/qdrant/storage |
||||
- ./snapshots:/qdrant/snapshots |
||||
environment: |
||||
- QDRANT__SERVICE__HTTP_PORT=6333 |
||||
- QDRANT__LOG_LEVEL=INFO |
||||
restart: unless-stopped |
||||
healthcheck: |
||||
test: ["CMD", "wget", "-q", "--spider", "http://localhost:6333/"] |
||||
interval: 30s |
||||
timeout: 10s |
||||
retries: 3 |
||||
deploy: |
||||
resources: |
||||
limits: |
||||
memory: 2G |
||||
cpus: '1.0' |
||||
|
||||
dozzle: |
||||
image: amir20/dozzle:latest |
||||
ports: |
||||
- "127.0.0.1:9999:8080" |
||||
volumes: |
||||
- /var/run/docker.sock:/var/run/docker.sock |
||||
restart: unless-stopped |
||||
``` |
||||
|
||||
### 3.2 mem0 配置 |
||||
|
||||
文件：`/root/.openclaw/workspace/skills/mem0-integration/config.yaml` |
||||
|
||||
```yaml |
||||
local: |
||||
vector_store: |
||||
provider: qdrant |
||||
config: |
||||
host: localhost |
||||
port: 6333 |
||||
collection_name: mem0_local |
||||
|
||||
llm: |
||||
provider: openai |
||||
config: |
||||
model: qwen-plus |
||||
api_base: https://dashscope.aliyuncs.com/compatible-mode/v1 |
||||
|
||||
embedder: |
||||
provider: openai |
||||
config: |
||||
model: text-embedding-v3 |
||||
api_base: https://dashscope.aliyuncs.com/compatible-mode/v1 |
||||
|
||||
master: |
||||
vector_store: |
||||
provider: qdrant |
||||
config: |
||||
host: 100.115.94.1 # Tailscale IP |
||||
port: 6333 |
||||
collection_name: mem0_shared |
||||
``` |
||||
|
||||
### 3.3 环境变量 |
||||
|
||||
文件：`/opt/mem0-center/.env` |
||||
|
||||
```bash |
||||
TS_AUTHKEY=tskey-auth-xxx |
||||
QDRANT_PORT=6333 |
||||
MEM0_DASHSCOPE_API_KEY=sk-xxx |
||||
MEM0_LLM_MODEL=qwen-plus |
||||
BUSINESS_TYPE=general |
||||
NODE_ROLE=center |
||||
NODE_NAME=mem0-general-center |
||||
``` |
||||
|
||||
--- |
||||
|
||||
## 4. 使用指南 |
||||
|
||||
### 4.1 部署流程 |
||||
|
||||
**步骤 1：系统检查** |
||||
```bash |
||||
./scripts/01-system-check.sh |
||||
``` |
||||
|
||||
**步骤 2：安装 Tailscale** |
||||
```bash |
||||
./scripts/02-install-tailscale.sh |
||||
# 访问 https://login.tailscale.com/admin/machines 确认上线 |
||||
``` |
||||
|
||||
**步骤 3：创建目录** |
||||
```bash |
||||
./scripts/03-create-directories.sh |
||||
``` |
||||
|
||||
**步骤 4：启动服务** |
||||
```bash |
||||
./scripts/05-start-center.sh |
||||
# 验证：curl http://localhost:6333/ |
||||
``` |
||||
|
||||
**步骤 5：安装 mem0 Skill** |
||||
```bash |
||||
./scripts/06-create-mem0-skill.sh |
||||
./scripts/07-install-dependencies.sh |
||||
``` |
||||
|
||||
**步骤 6：测试** |
||||
```bash |
||||
python3 /root/.openclaw/workspace/skills/mem0-integration/mem0_client.py |
||||
``` |
||||
|
||||
### 4.2 日常管理 |
||||
|
||||
**查看服务状态**： |
||||
```bash |
||||
cd /opt/mem0-center && docker compose ps |
||||
``` |
||||
|
||||
**查看日志**： |
||||
```bash |
||||
# Dozzle Web UI |
||||
http://100.115.94.1:9999 |
||||
|
||||
# 命令行 |
||||
cd /opt/mem0-center && docker compose logs -f |
||||
``` |
||||
|
||||
**监控系统**： |
||||
```bash |
||||
./scripts/12-monitoring.sh |
||||
``` |
||||
|
||||
**创建备份**： |
||||
```bash |
||||
./scripts/10-create-backup.sh |
||||
``` |
||||
|
||||
### 4.3 OpenClaw 命令 |
||||
|
||||
通过 Telegram 使用： |
||||
|
||||
``` |
||||
/memory add <内容> # 添加记忆 |
||||
/memory search <关键词> # 搜索记忆 |
||||
/memory list # 列出所有记忆 |
||||
/memory delete <ID> # 删除记忆 |
||||
/memory status # 查看状态 |
||||
``` |
||||
|
||||
--- |
||||
|
||||
## 5. 故障排除 |
||||
|
||||
### 5.1 Qdrant 无法启动 |
||||
|
||||
**症状**：`docker compose ps` 显示 qdrant-master 为 unhealthy |
||||
|
||||
**解决**： |
||||
```bash |
||||
# 查看日志 |
||||
docker compose logs qdrant-master |
||||
|
||||
# 检查端口占用 |
||||
netstat -tlnp | grep 6333 |
||||
|
||||
# 重启服务 |
||||
docker compose restart qdrant-master |
||||
``` |
||||
|
||||
### 5.2 mem0 初始化失败 |
||||
|
||||
**症状**：`mem0_client.py` 报错 |
||||
|
||||
**解决**： |
||||
```bash |
||||
# 检查环境变量 |
||||
export OPENAI_API_KEY="sk-xxx" |
||||
export OPENAI_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode/v1" |
||||
|
||||
# 测试导入 |
||||
python3 -c "from mem0 import Memory; print('OK')" |
||||
|
||||
# 检查 Qdrant 连接 |
||||
curl http://localhost:6333/ |
||||
``` |
||||
|
||||
### 5.3 Tailscale 断开 |
||||
|
||||
**症状**：`tailscale status` 显示 offline |
||||
|
||||
**解决**： |
||||
```bash |
||||
# 重新认证 |
||||
tailscale up --authkey=tskey-auth-xxx |
||||
|
||||
# 检查服务状态 |
||||
systemctl status tailscaled |
||||
``` |
||||
|
||||
--- |
||||
|
||||
## 6. 扩展部署 |
||||
|
||||
### 6.1 部署新 Agent 节点 |
||||
|
||||
**在新节点执行**： |
||||
|
||||
1. 安装 Tailscale |
||||
```bash |
||||
curl -fsSL https://tailscale.com/install.sh | sh |
||||
tailscale up --authkey=tskey-auth-xxx --hostname=mem0-general-agent-01 |
||||
``` |
||||
|
||||
2. 部署 Agent（文档待补充） |
||||
|
||||
### 6.2 业务类型扩展 |
||||
|
||||
命名规范： |
||||
- `mem0-general-center` - 通用业务（当前） |
||||
- `mem0-crypto-center` - 加密货币业务 |
||||
- `mem0-advert-center` - 广告业务 |
||||
- `mem0-life-center` - 生活业务 |
||||
|
||||
--- |
||||
|
||||
## 7. 安全配置 |
||||
|
||||
### 7.1 访问控制 |
||||
|
||||
- **Tailscale 内网**：所有服务仅通过 Tailscale IP 访问 |
||||
- **端口绑定**：仅绑定 127.0.0.1，不暴露公网 |
||||
- **API Key 管理**：存储在 .env 文件，权限 600 |
||||
|
||||
### 7.2 数据加密 |
||||
|
||||
- **Tailscale**：端到端加密 |
||||
- **Qdrant**：支持 TLS（可选配置） |
||||
- **备份**：建议加密存储 |
||||
|
||||
--- |
||||
|
||||
## 8. 性能优化 |
||||
|
||||
### 8.1 资源限制 |
||||
|
||||
```yaml |
||||
# Docker Compose |
||||
deploy: |
||||
resources: |
||||
limits: |
||||
memory: 2G |
||||
cpus: '1.0' |
||||
``` |
||||
|
||||
### 8.2 缓存配置 |
||||
|
||||
```yaml |
||||
# mem0 config |
||||
cache: |
||||
enabled: true |
||||
ttl: 300 # 5 分钟 |
||||
max_size: 1000 |
||||
``` |
||||
|
||||
--- |
||||
|
||||
## 9. 变更日志 |
||||
|
||||
### v1.0.0 (2026-02-22) |
||||
- ✅ 初始部署 |
||||
- ✅ 中心节点搭建完成 |
||||
- ✅ mem0 Integration 完成 |
||||
- ✅ 文档创建 |
||||
|
||||
--- |
||||
|
||||
## 10. 联系方式 |
||||
|
||||
- **Tailscale 管理**：https://login.tailscale.com/admin |
||||
- **mem0 文档**：https://docs.mem0.ai |
||||
- **Qdrant 文档**：https://qdrant.tech/documentation |
||||
- **OpenClaw 文档**：/root/.openclaw/workspace/docs |
||||
@ -0,0 +1,126 @@ |
||||
# mem0 记忆系统部署文档 |
||||
|
||||
## 快速开始 |
||||
|
||||
### 1. 系统检查 |
||||
```bash |
||||
./scripts/01-system-check.sh |
||||
``` |
||||
|
||||
### 2. 安装 Tailscale |
||||
```bash |
||||
./scripts/02-install-tailscale.sh |
||||
``` |
||||
|
||||
### 3. 创建目录 |
||||
```bash |
||||
./scripts/03-create-directories.sh |
||||
``` |
||||
|
||||
### 4. 启动中心服务 |
||||
```bash |
||||
./scripts/05-start-center.sh |
||||
``` |
||||
|
||||
### 5. 安装 mem0 Skill |
||||
```bash |
||||
./scripts/06-create-mem0-skill.sh |
||||
./scripts/07-install-dependencies.sh |
||||
``` |
||||
|
||||
### 6. 测试 |
||||
```bash |
||||
python3 /root/.openclaw/workspace/skills/mem0-integration/mem0_client.py |
||||
``` |
||||
|
||||
## 服务信息 |
||||
|
||||
### 中心节点 (vps-vaym) |
||||
- **Tailscale IP**: 100.115.94.1 |
||||
- **节点名称**: mem0-general-center |
||||
- **业务类型**: general |
||||
|
||||
### 服务端口 |
||||
| 服务 | 端口 | 访问方式 | |
||||
|------|------|---------| |
||||
| Qdrant | 6333 | Tailscale 内网 | |
||||
| Dozzle | 9999 | Tailscale 内网 | |
||||
|
||||
## 管理命令 |
||||
|
||||
### 查看服务状态 |
||||
```bash |
||||
cd /opt/mem0-center && docker compose ps |
||||
``` |
||||
|
||||
### 查看日志 |
||||
```bash |
||||
cd /opt/mem0-center && docker compose logs -f |
||||
``` |
||||
|
||||
### 重启服务 |
||||
```bash |
||||
cd /opt/mem0-center && docker compose restart |
||||
``` |
||||
|
||||
### 停止服务 |
||||
```bash |
||||
cd /opt/mem0-center && docker compose down |
||||
``` |
||||
|
||||
## 监控 |
||||
|
||||
### 运行监控脚本 |
||||
```bash |
||||
./scripts/12-monitoring.sh |
||||
``` |
||||
|
||||
### 查看 Dozzle 日志 |
||||
访问：http://100.115.94.1:9999 |
||||
|
||||
## 备份 |
||||
|
||||
### 创建备份 |
||||
```bash |
||||
./scripts/10-create-backup.sh |
||||
``` |
||||
|
||||
### 恢复备份 |
||||
```bash |
||||
cd /root/.openclaw/workspace/backup |
||||
tar -xzf backup-YYYYMMDD-HHMMSS.tar.gz |
||||
``` |
||||
|
||||
## 故障排除 |
||||
|
||||
### Qdrant 无法启动 |
||||
```bash |
||||
cd /opt/mem0-center |
||||
docker compose logs qdrant-master |
||||
``` |
||||
|
||||
### mem0 初始化失败 |
||||
```bash |
||||
export OPENAI_API_KEY="your-api-key" |
||||
export OPENAI_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode/v1" |
||||
python3 -c "from mem0 import Memory; print('OK')" |
||||
``` |
||||
|
||||
### Tailscale 断开 |
||||
```bash |
||||
tailscale status |
||||
tailscale up --authkey=YOUR_AUTH_KEY |
||||
``` |
||||
|
||||
## 下一步：部署 Agent 节点 |
||||
|
||||
1. 在新节点安装 Tailscale |
||||
2. 使用相同 Auth Key 认证 |
||||
3. 部署 Agent Docker Compose |
||||
4. 配置连接到中心 Qdrant (100.115.94.1) |
||||
|
||||
## 联系支持 |
||||
|
||||
- Tailscale 管理：https://login.tailscale.com/admin |
||||
- mem0 文档：https://docs.mem0.ai |
||||
- Qdrant 文档：https://qdrant.tech/documentation |
||||
@ -0,0 +1,5 @@ |
||||
[2026-02-21T06:17:19.888Z] [INFO] Graceful shutdown initiated |
||||
[2026-02-21T06:17:20.297Z] [INFO] Agent Health Monitor initialized |
||||
[2026-02-21T06:17:20.304Z] [INFO] Agent Health Monitor starting... |
||||
[2026-02-21T06:17:20.304Z] [INFO] Starting OpenClaw Gateway monitoring... |
||||
[2026-02-21T06:17:20.305Z] [INFO] Monitor is now active. Press Ctrl+C to stop. |
||||
@ -0,0 +1,101 @@ |
||||
#!/bin/bash |
||||
# /root/.openclaw/workspace/scripts/03-create-directories.sh |
||||
|
||||
set -e |
||||
|
||||
echo "📁 创建目录结构..." |
||||
|
||||
# 中心服务目录 |
||||
echo "创建中心服务目录..." |
||||
mkdir -p /opt/mem0-center/{qdrant_storage,snapshots,tailscale,logs,backup} |
||||
mkdir -p /opt/mem0-center/config |
||||
|
||||
# OpenClaw 目录检查 |
||||
echo "检查 OpenClaw 目录..." |
||||
if [ ! -d "/root/.openclaw/workspace" ]; then |
||||
echo "❌ OpenClaw workspace 不存在" |
||||
exit 1 |
||||
fi |
||||
echo "✅ OpenClaw workspace 已存在" |
||||
|
||||
# 脚本目录 |
||||
mkdir -p /root/.openclaw/workspace/scripts |
||||
mkdir -p /root/.openclaw/workspace/backup |
||||
mkdir -p /root/.openclaw/workspace/docs |
||||
|
||||
# mem0 Skill 目录（预创建） |
||||
mkdir -p /root/.openclaw/workspace/skills/mem0-integration |
||||
|
||||
# 设置权限 |
||||
chmod 755 /opt/mem0-center |
||||
chmod 755 /root/.openclaw/workspace/scripts |
||||
chmod 700 /opt/mem0-center/backup # 备份目录限制访问 |
||||
|
||||
# 创建目录说明文件 |
||||
cat > /opt/mem0-center/README.md << 'EOF' |
||||
# mem0-center - 中心节点 |
||||
|
||||
## 目录结构 |
||||
|
||||
- `qdrant_storage/` - Qdrant 向量数据库存储 |
||||
- `snapshots/` - Qdrant 快照备份 |
||||
- `tailscale/` - Tailscale 状态文件 |
||||
- `logs/` - 服务日志 |
||||
- `backup/` - 配置和数据备份 |
||||
- `config/` - 配置文件 |
||||
|
||||
## 服务 |
||||
|
||||
- Qdrant Master: 端口 6333 |
||||
- Dozzle (日志): 端口 9999 |
||||
- mem0 Server: 端口 8000 (可选) |
||||
|
||||
## 管理命令 |
||||
|
||||
```bash |
||||
# 启动服务 |
||||
docker compose up -d |
||||
|
||||
# 停止服务 |
||||
docker compose down |
||||
|
||||
# 查看状态 |
||||
docker compose ps |
||||
|
||||
# 查看日志 |
||||
docker compose logs -f |
||||
|
||||
# 重启服务 |
||||
docker compose restart |
||||
``` |
||||
|
||||
## Tailscale 信息 |
||||
|
||||
- 节点名称：mem0-general-center |
||||
- 业务类型：general |
||||
- 节点角色：center |
||||
- Tailscale IP: 100.115.94.1 |
||||
|
||||
## 访问方式 |
||||
|
||||
- Qdrant API: http://100.115.94.1:6333 |
||||
- Dozzle 日志：http://100.115.94.1:9999 |
||||
- mem0 API: http://100.115.94.1:8000 |
||||
EOF |
||||
|
||||
echo "" |
||||
echo "📊 目录结构:" |
||||
tree -L 2 /opt/mem0-center 2>/dev/null || ls -la /opt/mem0-center |
||||
|
||||
echo "" |
||||
echo "✅ 目录结构创建完成" |
||||
echo "" |
||||
echo "📁 已创建的目录:" |
||||
echo " /opt/mem0-center/qdrant_storage/ - Qdrant 数据存储" |
||||
echo " /opt/mem0-center/snapshots/ - Qdrant 快照备份" |
||||
echo " /opt/mem0-center/tailscale/ - Tailscale 状态" |
||||
echo " /opt/mem0-center/logs/ - 服务日志" |
||||
echo " /opt/mem0-center/backup/ - 备份文件" |
||||
echo " /opt/mem0-center/config/ - 配置文件" |
||||
echo "" |
||||
echo "📝 说明文档：/opt/mem0-center/README.md" |
||||
@ -0,0 +1,26 @@ |
||||
#!/bin/bash |
||||
# /root/.openclaw/workspace/scripts/07-install-dependencies.sh |
||||
|
||||
set -e |
||||
|
||||
echo "📦 安装依赖..." |
||||
|
||||
# 安装 mem0ai |
||||
echo "安装 mem0ai..." |
||||
pip install mem0ai --break-system-packages --quiet |
||||
|
||||
# 验证安装 |
||||
if python3 -c "import mem0" 2>/dev/null; then |
||||
echo "✅ mem0ai 安装成功" |
||||
else |
||||
echo "❌ mem0ai 安装失败" |
||||
exit 1 |
||||
fi |
||||
|
||||
# 安装 PyYAML |
||||
echo "安装 PyYAML..." |
||||
pip install pyyaml --break-system-packages --quiet |
||||
echo "✅ PyYAML 安装成功" |
||||
|
||||
echo "" |
||||
echo "✅ 依赖安装完成" |
||||
@ -0,0 +1,40 @@ |
||||
#!/bin/bash |
||||
# /root/.openclaw/workspace/scripts/10-create-backup.sh |
||||
|
||||
set -e |
||||
|
||||
echo "💾 创建备份..." |
||||
|
||||
BACKUP_DIR="/root/.openclaw/workspace/backup" |
||||
TIMESTAMP=$(date +%Y%m%d-%H%M%S) |
||||
BACKUP_PATH="$BACKUP_DIR/backup-$TIMESTAMP" |
||||
|
||||
mkdir -p "$BACKUP_PATH" |
||||
|
||||
# 备份 mem0 配置 |
||||
echo "📁 备份 mem0 配置..." |
||||
cp -r /root/.openclaw/workspace/skills/mem0-integration "$BACKUP_PATH/" 2>/dev/null || true |
||||
|
||||
# 备份中心服务配置 |
||||
echo "📁 备份中心服务配置..." |
||||
cp /opt/mem0-center/docker-compose.yml "$BACKUP_PATH/" 2>/dev/null || true |
||||
cp /opt/mem0-center/.env "$BACKUP_PATH/" 2>/dev/null || true |
||||
|
||||
# 创建 Qdrant 快照 |
||||
echo "📁 创建 Qdrant 快照..." |
||||
SNAPSHOT_RESPONSE=$(curl -s -X POST http://localhost:6333/collections/mem0_test/snapshots 2>/dev/null || echo '{"error":"collection not found"}') |
||||
echo " Qdrant 快照：$SNAPSHOT_RESPONSE" |
||||
|
||||
# 压缩备份 |
||||
cd "$BACKUP_DIR" |
||||
tar -czf "backup-$TIMESTAMP.tar.gz" "backup-$TIMESTAMP" |
||||
rm -rf "backup-$TIMESTAMP" |
||||
|
||||
echo "✅ 备份完成：$BACKUP_DIR/backup-$TIMESTAMP.tar.gz" |
||||
|
||||
# 保留最近 10 个备份 |
||||
ls -t "$BACKUP_DIR"/backup-*.tar.gz | tail -n +11 | xargs rm -f 2>/dev/null || true |
||||
|
||||
echo "" |
||||
echo "📊 当前备份:" |
||||
ls -lht "$BACKUP_DIR"/backup-*.tar.gz | head -5 |
||||
@ -0,0 +1,41 @@ |
||||
#!/bin/bash |
||||
# /root/.openclaw/workspace/scripts/12-monitoring.sh |
||||
|
||||
set -e |
||||
|
||||
echo "📊 系统监控..." |
||||
echo "=" * 60 |
||||
|
||||
# 服务状态 |
||||
echo -e "\n=== 服务状态 ===" |
||||
cd /opt/mem0-center && docker compose ps |
||||
|
||||
# Qdrant 健康 |
||||
echo -e "\n=== Qdrant 健康 ===" |
||||
curl -s http://localhost:6333/ | python3 -m json.tool |
||||
|
||||
# 内存使用 |
||||
echo -e "\n=== 内存使用 ===" |
||||
free -h |
||||
|
||||
# 磁盘使用 |
||||
echo -e "\n=== 磁盘使用 ===" |
||||
df -h /opt/mem0-center |
||||
|
||||
# Tailscale 状态 |
||||
echo -e "\n=== Tailscale 状态 ===" |
||||
tailscale status 2>/dev/null | grep mem0-general-center || echo "Tailscale 未运行" |
||||
|
||||
# mem0 状态 |
||||
echo -e "\n=== mem0 状态 ===" |
||||
cd /root/.openclaw/workspace/skills/mem0-integration && python3 -c " |
||||
from mem0_client import Mem0Client |
||||
client = Mem0Client() |
||||
status = client.get_status() |
||||
print(f'初始化：{\"✅\" if status[\"initialized\"] else \"❌\"}') |
||||
print(f'Qdrant: {status[\"qdrant\"]}') |
||||
" |
||||
|
||||
echo -e "\n" |
||||
echo "=" * 60 |
||||
echo "✅ 监控完成" |
||||
@ -0,0 +1,35 @@ |
||||
# mem0-integration Skill |
||||
|
||||
## 功能说明 |
||||
|
||||
集成 mem0 记忆系统，为 OpenClaw 提供： |
||||
- ✅ 本地记忆存储（Qdrant Local） |
||||
- ✅ 共享记忆同步（Qdrant Master） |
||||
- ✅ 语义搜索 |
||||
- ✅ 多 Agent 协作 |
||||
- ✅ 分层记忆管理 |
||||
|
||||
## 架构 |
||||
|
||||
``` |
||||
Agent → mem0 Client → Qdrant Local → (异步同步) → Qdrant Master (100.115.94.1) |
||||
``` |
||||
|
||||
## 配置 |
||||
|
||||
编辑 `/root/.openclaw/workspace/skills/mem0-integration/config.yaml` |
||||
|
||||
## 命令 |
||||
|
||||
- `/memory add <内容>` - 添加记忆 |
||||
- `/memory search <关键词>` - 搜索记忆 |
||||
- `/memory list` - 列出所有记忆 |
||||
- `/memory delete <ID>` - 删除记忆 |
||||
- `/memory sync` - 手动同步到中心 |
||||
- `/memory status` - 查看状态 |
||||
|
||||
## 依赖 |
||||
|
||||
- mem0ai (pip install mem0ai) |
||||
- Qdrant (Docker) |
||||
- pyyaml |
||||
Binary file not shown.
@ -0,0 +1,79 @@ |
||||
#!/usr/bin/env python3 |
||||
""" |
||||
mem0 Commands for OpenClaw |
||||
""" |
||||
|
||||
from .mem0_client import Mem0Client |
||||
|
||||
mem0 = Mem0Client() |
||||
|
||||
async def handle_memory_command(args: str, context: dict) -> str: |
||||
"""处理 /memory 命令""" |
||||
parts = args.split(' ', 2) |
||||
action = parts[0] if parts else 'help' |
||||
|
||||
user_id = context.get('user_id', 'default') |
||||
agent_id = context.get('agent_id', 'main') |
||||
|
||||
if action == 'add': |
||||
content = parts[1] if len(parts) > 1 else '' |
||||
return await handle_add_memory(content, user_id, agent_id) |
||||
elif action == 'search': |
||||
query = parts[1] if len(parts) > 1 else '' |
||||
return await handle_search_memory(query, user_id, agent_id) |
||||
elif action == 'list': |
||||
return await handle_list_memories(user_id, agent_id) |
||||
elif action == 'delete': |
||||
memory_id = parts[1] if len(parts) > 1 else '' |
||||
return await handle_delete_memory(memory_id, user_id, agent_id) |
||||
elif action == 'sync': |
||||
return await handle_sync_memory(user_id, agent_id) |
||||
elif action == 'status': |
||||
return handle_status(user_id, agent_id) |
||||
else: |
||||
return get_help_text() |
||||
|
||||
async def handle_add_memory(content: str, user_id: str, agent_id: str) -> str: |
||||
if not content: |
||||
return "❌ 请提供记忆内容" |
||||
result = mem0.add([{"role": "user", "content": content}], user_id=user_id, agent_id=agent_id) |
||||
if 'error' in result: |
||||
return f"❌ 添加失败：{result['error']}" |
||||
return f"✅ 记忆已添加：{content[:100]}..." |
||||
|
||||
async def handle_search_memory(query: str, user_id: str, agent_id: str) -> str: |
||||
if not query: |
||||
return "❌ 请提供关键词" |
||||
results = mem0.search(query, user_id=user_id, agent_id=agent_id, limit=5) |
||||
if not results: |
||||
return f"🔍 未找到相关记忆" |
||||
response = f"🔍 找到 {len(results)} 条记忆:\n\n" |
||||
for i, mem in enumerate(results, 1): |
||||
response += f"{i}. {mem.get('memory', 'N/A')}\n" |
||||
return response |
||||
|
||||
async def handle_list_memories(user_id: str, agent_id: str) -> str: |
||||
results = mem0.get_all(user_id=user_id, agent_id=agent_id) |
||||
if not results: |
||||
return "📭 暂无记忆" |
||||
return f"📋 共 {len(results)} 条记忆" |
||||
|
||||
async def handle_delete_memory(memory_id: str, user_id: str, agent_id: str) -> str: |
||||
if not memory_id: |
||||
return "❌ 请提供 ID" |
||||
success = mem0.delete(memory_id, user_id=user_id, agent_id=agent_id) |
||||
return "✅ 已删除" if success else "❌ 删除失败" |
||||
|
||||
async def handle_sync_memory(user_id: str, agent_id: str) -> str: |
||||
return "🔄 同步功能开发中" |
||||
|
||||
def handle_status(user_id: str, agent_id: str) -> str: |
||||
status = mem0.get_status() |
||||
response = "📊 mem0 状态:\n" |
||||
response += f"本地：{'✅' if status['local_initialized'] else '❌'}\n" |
||||
response += f"共享：{'✅' if status['master_initialized'] else '❌'}\n" |
||||
return response |
||||
|
||||
def get_help_text() -> str: |
||||
return """📖 用法：/memory <命令> |
||||
命令：add, search, list, delete, sync, status""" |
||||
@ -0,0 +1,46 @@ |
||||
# mem0 Integration Configuration |
||||
|
||||
# 本地 Qdrant 配置 |
||||
local: |
||||
vector_store: |
||||
provider: qdrant |
||||
config: |
||||
host: localhost |
||||
port: 6333 |
||||
collection_name: mem0_local |
||||
|
||||
llm: |
||||
provider: openai |
||||
config: |
||||
model: qwen-plus |
||||
api_base: https://dashscope.aliyuncs.com/compatible-mode/v1 |
||||
api_key: sk-c1715ee0479841399fd359c574647648 |
||||
|
||||
embedder: |
||||
provider: openai |
||||
config: |
||||
model: text-embedding-v3 |
||||
api_base: https://dashscope.aliyuncs.com/compatible-mode/v1 |
||||
api_key: sk-c1715ee0479841399fd359c574647648 |
||||
|
||||
# 中心 Qdrant 配置（共享记忆） |
||||
master: |
||||
vector_store: |
||||
provider: qdrant |
||||
config: |
||||
host: 100.115.94.1 |
||||
port: 6333 |
||||
collection_name: mem0_shared |
||||
|
||||
# 同步配置 |
||||
sync: |
||||
enabled: true |
||||
interval: 300 |
||||
batch_size: 50 |
||||
retry_attempts: 3 |
||||
|
||||
# 缓存配置 |
||||
cache: |
||||
enabled: true |
||||
ttl: 300 |
||||
max_size: 1000 |
||||
@ -0,0 +1,150 @@ |
||||
#!/usr/bin/env python3 |
||||
""" |
||||
mem0 OpenClaw Commands |
||||
处理 /memory 命令 |
||||
""" |
||||
|
||||
import os |
||||
import sys |
||||
|
||||
# 设置环境变量 |
||||
os.environ['OPENAI_API_BASE'] = 'https://dashscope.aliyuncs.com/compatible-mode/v1' |
||||
os.environ['OPENAI_API_KEY'] = 'sk-c1715ee0479841399fd359c574647648' |
||||
|
||||
sys.path.insert(0, os.path.dirname(__file__)) |
||||
|
||||
from mem0_client import Mem0Client |
||||
|
||||
# 初始化客户端 |
||||
mem0 = Mem0Client() |
||||
|
||||
def handle_memory_command(args: str, context: dict) -> str: |
||||
""" |
||||
处理 /memory 命令 |
||||
|
||||
Args: |
||||
args: 命令参数 |
||||
context: 上下文信息 (user_id, agent_id 等) |
||||
|
||||
Returns: |
||||
回复文本 |
||||
""" |
||||
parts = args.split(' ', 2) |
||||
action = parts[0] if parts else 'help' |
||||
|
||||
user_id = context.get('user_id', 'default') |
||||
agent_id = context.get('agent_id', 'main') |
||||
|
||||
if action == 'add': |
||||
content = parts[1] if len(parts) > 1 else '' |
||||
return handle_add(content, user_id, agent_id) |
||||
elif action == 'search': |
||||
query = parts[1] if len(parts) > 1 else '' |
||||
return handle_search(query, user_id, agent_id) |
||||
elif action == 'list': |
||||
return handle_list(user_id, agent_id) |
||||
elif action == 'delete': |
||||
memory_id = parts[1] if len(parts) > 1 else '' |
||||
return handle_delete(memory_id, user_id, agent_id) |
||||
elif action == 'status': |
||||
return handle_status() |
||||
else: |
||||
return get_help() |
||||
|
||||
def handle_add(content: str, user_id: str, agent_id: str) -> str: |
||||
"""添加记忆""" |
||||
if not content: |
||||
return "❌ 请提供记忆内容\n\n用法：/memory add <内容>" |
||||
|
||||
result = mem0.add([{"role": "user", "content": content}], user_id=user_id) |
||||
|
||||
if result and 'error' in result: |
||||
return f"❌ 添加失败：{result['error']}" |
||||
|
||||
return f"✅ 记忆已添加\n\n📝 内容：{content[:200]}..." |
||||
|
||||
def handle_search(query: str, user_id: str, agent_id: str) -> str: |
||||
"""搜索记忆""" |
||||
if not query: |
||||
return "❌ 请提供搜索关键词\n\n用法：/memory search <关键词>" |
||||
|
||||
results = mem0.search(query, user_id=user_id, limit=5) |
||||
|
||||
if not results: |
||||
return f"🔍 未找到与 \"{query}\" 相关的记忆" |
||||
|
||||
response = f"🔍 找到 {len(results)} 条相关记忆:\n\n" |
||||
for i, mem in enumerate(results, 1): |
||||
memory_text = mem.get('memory', 'N/A') |
||||
response += f"{i}. {memory_text}\n" |
||||
if mem.get('created_at'): |
||||
response += f" 📅 {mem['created_at']}\n" |
||||
response += "\n" |
||||
|
||||
return response |
||||
|
||||
def handle_list(user_id: str, agent_id: str) -> str: |
||||
"""列出所有记忆""" |
||||
results = mem0.get_all(user_id=user_id) |
||||
|
||||
if not results: |
||||
return "📭 暂无记忆" |
||||
|
||||
response = f"📋 共 {len(results)} 条记忆:\n\n" |
||||
for i, mem in enumerate(results[:20], 1): |
||||
memory_text = mem.get('memory', 'N/A') |
||||
truncated = memory_text[:100] + '...' if len(memory_text) > 100 else memory_text |
||||
response += f"{i}. {truncated}\n" |
||||
response += f" 🆔 {mem.get('id', 'N/A')}\n\n" |
||||
|
||||
if len(results) > 20: |
||||
response += f"... 还有 {len(results) - 20} 条记忆\n" |
||||
|
||||
return response |
||||
|
||||
def handle_delete(memory_id: str, user_id: str, agent_id: str) -> str: |
||||
"""删除记忆""" |
||||
if not memory_id: |
||||
return "❌ 请提供记忆 ID\n\n用法：/memory delete <ID>" |
||||
|
||||
success = mem0.delete(memory_id, user_id=user_id) |
||||
|
||||
if success: |
||||
return f"✅ 记忆已删除：{memory_id}" |
||||
else: |
||||
return f"❌ 删除失败：记忆不存在或权限不足" |
||||
|
||||
def handle_status() -> str: |
||||
"""查看状态""" |
||||
status = mem0.get_status() |
||||
|
||||
response = "📊 mem0 状态:\n\n" |
||||
response += f"本地记忆：{'✅' if status['initialized'] else '❌'}\n" |
||||
response += f"Qdrant 地址：{status['qdrant']}\n" |
||||
|
||||
return response |
||||
|
||||
def get_help() -> str: |
||||
"""获取帮助文本""" |
||||
return """📖 mem0 记忆管理命令 |
||||
|
||||
用法：/memory <命令> [参数] |
||||
|
||||
可用命令: |
||||
add <内容> - 添加记忆 |
||||
search <关键词> - 搜索记忆 |
||||
list - 列出所有记忆 |
||||
delete <ID> - 删除记忆 |
||||
status - 查看状态 |
||||
help - 显示帮助 |
||||
|
||||
示例: |
||||
/memory add 用户偏好使用 UTC 时区 |
||||
/memory search 时区 |
||||
/memory list |
||||
/memory status""" |
||||
|
||||
# 测试 |
||||
if __name__ == '__main__': |
||||
print("测试 /memory 命令...") |
||||
print(handle_status()) |
||||
@ -0,0 +1,36 @@ |
||||
{ |
||||
"name": "mem0-integration", |
||||
"version": "1.0.0", |
||||
"description": "mem0 记忆系统集成，提供长期记忆存储和语义搜索", |
||||
"author": "OpenClaw Team", |
||||
"enabled": true, |
||||
"commands": [ |
||||
{ |
||||
"name": "memory", |
||||
"description": "记忆管理命令", |
||||
"handler": "openclaw_commands.handle_memory_command", |
||||
"usage": "/memory <add|search|list|delete|status|help> [参数]", |
||||
"examples": [ |
||||
"/memory add 用户偏好使用 UTC 时区", |
||||
"/memory search 时区", |
||||
"/memory list", |
||||
"/memory status" |
||||
] |
||||
} |
||||
], |
||||
"config": { |
||||
"qdrant": { |
||||
"host": "localhost", |
||||
"port": 6333, |
||||
"collection_name": "mem0_local" |
||||
}, |
||||
"llm": { |
||||
"model": "qwen-plus", |
||||
"api_base": "https://dashscope.aliyuncs.com/compatible-mode/v1" |
||||
} |
||||
}, |
||||
"dependencies": [ |
||||
"mem0ai", |
||||
"pyyaml" |
||||
] |
||||
} |
||||
Loading…
Reference in new issue