Operación Privada AI-LAB

Runbook operativo interno del laboratorio AI-LAB.

Documento interno operativo del laboratorio AI-LAB.

Infraestructura Principal

Section titled “Infraestructura Principal”
ServicioHostPuertoEstado
AI-LAB Main Node192.168.1.30-ONLINE
Main LM Studio192.168.1.2001234ONLINE
GPU Node RX9070XT192.168.1.501234ONLINE
GPU Node RX7900XT192.168.1.601234OFFLINE
Router API192.168.1.308008ONLINE
OpenCode192.168.1.304096ONLINE
Traefik192.168.1.3080/443ONLINE
Qdrant192.168.1.306333ONLINE
Open WebUI192.168.1.303000ONLINE
Portainer192.168.1.309443ONLINE

Router Cognitivo

Section titled “Router Cognitivo”

Estado

Section titled “Estado”

Servicio:

Terminal window
sudo systemctl status ialab-router-api

Restart:

Terminal window
sudo systemctl restart ialab-router-api

Logs:

Terminal window
journalctl -u ialab-router-api -f

Health:

Terminal window
curl http://127.0.0.1:8008/health

Modelos:

Terminal window
curl http://127.0.0.1:8008/v1/models

Test chat:

Terminal window
curl -s http://127.0.0.1:8008/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ailab-router/auto",
    "messages": [
      {
        "role": "user",
        "content": "di hola"
      }
    ]
  }'

OpenCode

Section titled “OpenCode”

Lanzar OpenCode

Section titled “Lanzar OpenCode”
Terminal window
/opt/ai-lab/runtime/opencode_ui.sh

Matar proceso

Section titled “Matar proceso”
Terminal window
pkill -f opencode

Ver proceso

Section titled “Ver proceso”
Terminal window
ps aux | grep opencode

Acceso web

Section titled “Acceso web”
http://192.168.1.30:4096

LM Studio

Section titled “LM Studio”

Endpoint REST

Section titled “Endpoint REST”
http://IP:1234

Ver modelos

Section titled “Ver modelos”
Terminal window
curl http://192.168.1.200:1234/v1/models

Test chat directo

Section titled “Test chat directo”
Terminal window
curl -s http://192.168.1.200:1234/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemma-4-e4b",
    "messages": [
      {
        "role": "user",
        "content": "hola"
      }
    ]
  }'

Snapshot Runtime

Section titled “Snapshot Runtime”

Generar snapshot

Section titled “Generar snapshot”
Terminal window
sudo systemctl restart ialab-live-state

Ver snapshot actual

Section titled “Ver snapshot actual”
Terminal window
cat /opt/ai-lab-data/snapshots/current/system_snapshot.json | jq

Ver estado LLM

Section titled “Ver estado LLM”
Terminal window
cat /opt/ai-lab-data/snapshots/current/system_snapshot.json | jq '.llm'

Docker

Section titled “Docker”

Contenedores

Section titled “Contenedores”
Terminal window
docker ps

Redes

Section titled “Redes”
Terminal window
docker network ls

Reiniciar Traefik

Section titled “Reiniciar Traefik”
Terminal window
docker restart traefik

Traefik

Section titled “Traefik”

Dashboard

Section titled “Dashboard”
http://192.168.1.30:8080

Configuración

Section titled “Configuración”
/opt/ai-lab/stacks/traefik/docker-compose.yml

Reiniciar stack

Section titled “Reiniciar stack”
Terminal window
cd /opt/ai-lab/stacks/traefik
docker compose up -d

Cloudflare Tunnel

Section titled “Cloudflare Tunnel”

Estado

Section titled “Estado”
Terminal window
systemctl status cloudflared

Logs

Section titled “Logs”
Terminal window
journalctl -u cloudflared -f

Astro / ialab-docs

Section titled “Astro / ialab-docs”

Desarrollo

Section titled “Desarrollo”
Terminal window
npm run dev -- --host

Build

Section titled “Build”
Terminal window
npm run build

Reiniciar docs

Section titled “Reiniciar docs”
Terminal window
sudo systemctl restart ialab-docs

Logs

Section titled “Logs”
Terminal window
journalctl -u ialab-docs -f

Git

Section titled “Git”

Snapshot rápido

Section titled “Snapshot rápido”
Terminal window
git add .
git commit -m "snapshot"

Estado

Section titled “Estado”
Terminal window
git status

Arquitectura Operativa

Section titled “Arquitectura Operativa”
flowchart LR

User --> Cloudflare
Cloudflare --> Tunnel
Tunnel --> Traefik

Traefik --> Astro
Traefik --> OpenCode
Traefik --> OpenWebUI
Traefik --> RouterAPI

RouterAPI --> LMStudioLocal
RouterAPI --> RX9070XT
RouterAPI --> RX7900XT

RouterAPI --> Qdrant

OpenCode --> RouterAPI
OpenWebUI --> RouterAPI

Estado Actual del LAB

Section titled “Estado Actual del LAB”
  • Main LM Studio operativo
  • Router OpenAI-compatible funcional
  • OpenCode operativo
  • Traefik + Cloudflare activos
  • Observabilidad viva
  • Arquitectura distribuida parcial
  • RX9070XT pendiente de estabilización
  • RX7900XT apagado

TODO

Section titled “TODO”
  • Failover inteligente real
  • Auto healthcheck por request
  • Métricas Prometheus
  • Grafana NOC
  • Routing cognitivo avanzado
  • Persistencia de memoria
  • Multi-agent runtime
  • Balanceo automático GPU