Monitoring

Monitoring

Zabbix-based infrastructure monitoring for HomePilot and Proxmox VE.

Overview

HomePilot Agent includes Zabbix monitoring for:

• 5 zabbix-agent2 hosts: database, monitoring, proxy, app-server, agent
• 3 Proxmox VE nodes: monitored via API (no SSH agent possible on PVE bare-metal)
• Docker containers: tracked per host
• CVE tracking: automated vulnerability scanning

Architecture

┌──────────────────┐
│   Zabbix Server   │  (PostgreSQL backend)
│   + Web UI        │
└────────┬─────────┘
         │
    ┌────┼────┬─────┬─────┐
    │    │    │     │     │
    ▼    ▼    ▼     ▼     ▼
  Host  Host  Host  Host  Host
  (db)  (mon) (proxy)(app) (agent)
         │
         ▼
    Proxmox API
    (3 PVE nodes)

Zabbix Setup

Server

1. Install Zabbix server with PostgreSQL backend
2. Configure zabbix-server.conf:

   DBHost=<database-host>
   DBName=zabbix
   DBUser=zabbix
   DBPassword=<password>

3. Install PHP PostgreSQL extension: apt install php8.3-pgsql
4. Generate locale: locale-gen en_US.UTF-8

Agent (on each managed host)

apt install zabbix-agent2
# /etc/zabbix/zabbix_agent2.conf
Server=<monitoring-host-ip>
ServerActive=<monitoring-host-ip>
systemctl restart zabbix-agent2

PVE Nodes (API-only)

Proxmox VE bare-metal hosts reject SSH (PermitRootLogin=no), so use HTTP agent monitoring:

1. Create a PVE API token (read-only, PVEAuditor role)
2. In Zabbix, add PVE hosts with HTTP agent items pointing to https://<pve-host>:8006/api2/json/
3. Use the PVE token in HTTP headers for authentication

Hosts

Host	Type	Monitoring
database-host	LXC	zabbix-agent2 (active)
monitoring-host	LXC	zabbix-agent2 (active)
proxy-host	LXC	zabbix-agent2 (active)
app-server	VM	zabbix-agent2 (active)
agent-host	VM	zabbix-agent2 (active)
pve1	Bare-metal	Proxmox API (HTTP agent)
pve2	Bare-metal	Proxmox API (HTTP agent)
pve3	Bare-metal	Proxmox API (HTTP agent)

Key Metrics

• CPU utilization per host
• Memory usage and availability
• Disk space and I/O
• Network traffic and errors
• Docker container status
• Proxmox VM/LXC status and resource usage
• PVE cluster health

Alerting

Zabbix triggers and actions can be configured to send alerts via:

• Email: Critical severity alerts
• Webhook: Real-time notifications to Matrix/Slack/Discord
• Scripts: Custom remediation actions

Known Limitations

• PVE bare-metal hosts cannot run zabbix-agent2 (SSH access restricted)
• PVE local-lvm at ~90% is normal for thin provisioning — monitor actual allocation
• Ollama embedding service uses an SSH reverse tunnel (ephemeral) — restart required after reboot

Screenshots

See Also

• Deployment — Docker Compose setup and configuration
• Security — Hardening and least-privilege setup