High-Availability Infrastructure

Multi-Server DevOps Architecture with Self-Healing Capabilities

Designed by

Đỗ Cao Hiếu

DevOps Engineer

Layer 1: Traffic Ingress
🌐
Internet
Users / Clients
→
☁️
Cloudflare
DNS + CDN + DDoS
→
⚖️
Load Balancer
xxx.xxx.xxx.182 • 25TB BW
Traefik :XX/:XXX
PG Proxy :XXXX | Redis :XXXX
Layer 2: Application Servers
Server 1 - Claw Đại
PG Leader
xxx.xxx.xxx.233 • VPN: 10.x.x.2
N8N Workflow (4 containers)
Chatwoot (3 containers)
+ Traefik, Exporters, ...
Native HA Services:
Patroni ⭐ Redis ⭐ etcd
Server 2 - Claw Đệ
🤖
xxx.xxx.xxx.246 • VPN: 10.x.x.3
Jenkins CI/CD
Harbor Registry (7 containers)
Semaphore (Ansible UI)
Terrakube (Terraform UI)
Cognee AI + Neo4j
+ RedisInsight, Traefik, ...
Native HA Services:
Patroni Redis etcd
Server 3 - Claw Út
🤖
xxx.xxx.xxx.221 • VPN: 10.x.x.1
Prometheus + Alertmanager
Grafana Dashboards
Loki + Tempo (Tracing)
Firecrawl (6 containers)
+ Stirling PDF, Traefik, ...
Native HA Services:
Patroni Redis etcd
Layer 3: Supporting Infrastructure
Oracle Server
20 containers
xxx.xxx.xxx.172 • VPN: 10.x.x.5 • 10TB BW
HashiCorp Vault
GitLab + Runner
Nextcloud + MinIO
Cognee MCP
Tmail (2 instances)
Rancher (K8s)
GWS + DCHighSchool
+ Traefik, ...
🔒 WireGuard Full Mesh VPN
Encrypted P2P connections between all nodes
10.x.x.1
S3
10.x.x.2
S1
10.x.x.3
S2
10.x.x.4
LB
10.x.x.5
Oracle
All nodes connected via WireGuard on port XXXXX
High Availability Components
🐘
Patroni Cluster
PostgreSQL HA
Leader: S1
Replicas: S2, S3
🔴
Redis Sentinel
Cache HA
Master: S1
Sentinels: 3 nodes
📦
etcd Cluster
Distributed KV Store
3-node cluster
S1, S2, S3
🔀
HAProxy
Local PG Routing
On each node
Auto leader discovery
Self-Healing AI Agent Mesh
🤖 OpenClaw Agent Mesh
Self-Healing Enabled
🦞 Claw Đại (S1)
Monitors: S2, S3
Can restart: Gateway on S2, S3
Port: XXXXX
🦞 Claw Đệ (S2)
Monitors: S1, S3
Can restart: Gateway on S1, S3
Port: XXXXX
🦞 Claw Út (S3)
Monitors: S1, S2
Can restart: Gateway on S1, S2
Port: XXXXX
Features: Leader Election • File Locking • SSH Auto-Recovery • Alertmanager Webhook Integration • DNS Failover
Technology Stack
Docker Traefik WireGuard Patroni Redis Sentinel etcd Prometheus Grafana Loki Harbor Jenkins GitLab Vault N8N Chatwoot OpenClaw AI Cloudflare

Infrastructure designed and maintained by Đỗ Cao Hiếu

5 Servers • 79 Containers • Full HA • Self-Healing AI Agents • Zero Downtime

Generated: February 2026