docs: layered self-healing architecture

This commit is contained in:
root committed 2026-09-28 16:54:40 +03:00
1 parent c43c60a9cb
commit f3e24bb7a8
1 file changed
+38
+38
View File
@@ -0,0 +1,38 @@
# Self-Healing Infrastructure (Layered)
## Слои
| L0 | Docker restart policy | ~1-2с | все контейнеры |
| L1 | Kuma→n8n→Remediation API | 60-120с | 11 whitelist |
| L2 | Gitea issue + Telegram | мин-часы | человек |
## Layer 0
Docker `restart: unless-stopped` поднимает контейнер за 1-2с.
Kuma часто не успевает заметить падение.
## Layer 1
Kuma (57 мониторов) → webhook → n8n (TSqu00NJx2JOrtoN, 10 нод) →
Remediation API (http://192.168.1.10:5000, Bearer auth, whitelist) →
pct exec docker restart → Telegram ✅/❌
## Компоненты
- Kuma: CT 107:3001, https://status.buryans.ru
- n8n: CT 108:5680 (прямой) / 5678 (oauth2-proxy)
- Remediation API: PVE host, systemd remediation-api.service
- /opt/remediation-api/remediation_api.py (stdlib)
- Токен: /opt/remediation-api/api_token.txt
## Whitelist (11 сервисов)
matrix-synapse, element-web, mautrix-telegram, matrix-coturn, livekit (CT 155)
vaultwarden, nextcloud (CT 109)
gitea (CT 140)
authentik (CT 114)
uptime-kuma, adguard (CT 107)
## Тест
curl -X POST http://192.168.1.108:5680/webhook/uptime-kuma -H "Content-Type: application/json" -d '{"monitorName":"Matrix Synapse","status":"0","shouldAlert":true}'
## Логи
- Remediation: journalctl -u remediation-api -f
- n8n: pct exec 108 -- docker logs n8n -f
Создано: 2026-09-28. Статус: ✅ активно.