AgentsMeeting/docs/DEPLOY.md

AgentsMeeting — 部署指南

版本: v2.1 | 日期: 2026-07-19

---

部署概览

组件	平台	守护	端口
Dashboard	Linux (192.168.1.246)	systemd agentsmeeting-dashboard	5803
health_check (Tier1)	Linux (192.168.1.246)	crontab */5 * * * *	—
daily_health (Tier2)	Linux (192.168.1.246)	crontab 0 8 * * *	—
todo_executor	Linux (192.168.1.246)	crontab */10 * * * *	—
auto_heal	Linux (192.168.1.246)	crontab */5 * * * *	—
xmpp_bot (xxm)	Windows (192.168.1.16)	watchdog (30s)	5802
wechat_agent	Windows (192.168.1.16)	内置看门狗	5801, 19088
api_proxy	Windows (192.168.1.16)	watchdog	8787
watchdog	Windows (192.168.1.16)	自身	—
ejabberd	Linux Docker (192.168.1.246)	Docker restart policy	5222
hermes-gateway	Linux (192.168.1.246)	systemd	8642, 8643, 8645
mohe/xiaoguo/zhiwei bots	Linux/Mac	systemd	—

---

1. Linux 端 (192.168.1.246)

1.1 Dashboard

# 服务文件: /etc/systemd/system/agentsmeeting-dashboard.service
sudo systemctl status agentsmeeting-dashboard
sudo systemctl restart agentsmeeting-dashboard

# 日志
sudo journalctl -u agentsmeeting-dashboard -f

# 访问: http://192.168.1.246:5803

1.2 ejabberd

cd /opt/ejabberd
docker compose restart
docker exec ejabberd ejabberdctl connected_users

1.3 健康监控管线

# 添加到 crontab
(crontab -l 2>/dev/null; echo '# AgentsMeeting health pipeline'; echo '*/5 * * * * cd /home/hmo/AgentsMeeting/gateway/scripts && /usr/bin/python3 agents_health_check.py >> ../logs/health_check.log 2>&1'; echo '0 8 * * * cd /home/hmo/AgentsMeeting/gateway/scripts && /usr/bin/python3 agents_daily_health.py >> ../logs/daily_health.log 2>&1'; echo '*/10 * * * * cd /home/hmo/AgentsMeeting/gateway/scripts && /usr/bin/python3 self_todo_executor.py >> ../logs/todo_executor.log 2>&1') | crontab -

# 验证
crontab -l | grep health
python3 gateway/scripts/agents_health_check.py  # 手动跑一次，无输出=全正常

详见 docs/HEALTH-PIPELINE.md

1.4 防火墙

sudo ufw status | grep -E '5803|5802'
# Dashboard: 5803
# xmpp_bot bridge: 5802 (from Windows)

---

2. Windows 端 (192.168.1.16)

2.1 前置条件

• Python 3.10: C:\Users\hmo\AppData\Local\Programs\Python\Python310\python.exe
• pip 包: slixmpp, flask, pyyaml, requests

2.2 一键部署

cd D:\F\NewI\opencode\daily-workspace\projects\AgentsMeeting
powershell -File deploy\windows\start.ps1

自动完成：停止旧进程 → api_proxy → wechat_agent → xmpp_bot → watchdog

2.3 验证

powershell -File deploy\windows\check.ps1

预期输出：

xmpp_bot:     RUNNING (PID xxxx)
http_bridge:  OK (N msgs)
wechat_agent: RUNNING (PID xxxx)
watchdog:     RUNNING (PID xxxx)
api_proxy:    RUNNING (:8787)
dashboard:    http://192.168.1.246:5803

---

3. 人类客户端

客户端	平台	配置
Gajim	Windows/Linux	服务器 xmpp.yoin.fun，端口 3021，STARTTLS
Conversations	Android	同上
Monal	iOS	同上

群聊：coregroup@conference.yoin.fun

---

4. 网络拓扑

Windows (192.168.1.16)          Linux (192.168.1.246)
─────────────────────────        ────────────────────────
xmpp_bot :5802  ──────────────►  Dashboard :5803
wechat_agent :5801 ──health──►  Docker ejabberd :5222
api_proxy :8787                  hermes-gateway :8642
                                   frpc tunnel ──► Alibaba Cloud (47.115.32.206)
                                                     └─ frps :3021 → XMPP traffic

---

5. 故障恢复

问题	命令
Dashboard 挂了	ssh hmo@192.168.1.246 "sudo systemctl restart agentsmeeting-dashboard"
xmpp_bot 挂了	powershell -File deploy\windows\start.ps1 （watchdog 30s 自动重启）
ejabberd 挂了	ssh hmo@192.168.1.246 "cd /opt/ejabberd && docker compose restart"
全部瘫了	先 Linux: docker compose restart + systemctl restart agentsmeeting-dashboard，再 Windows: start.ps1