NetworkObservability Stack — The Capes Homelab

Hybrid observability stack for the Mac Pro at 192.168.1.30, The Capes, Ventura CA. Monitors host system metrics, Docker services, Plex, SMB, Orbi mesh network, and all LAN devices.

Current version: 2.0.0

Stack: Prometheus · Grafana · Loki · Promtail · Telegraf · Alertmanager · Node Exporter · cAdvisor

---

Architecture

Orbi RBR750 Router (192.168.1.1)
  │  SOAP API ←── telegraf (exec: orbi_metrics.sh every 30s)
  │               per-device RSSI, linkspeed, mesh node, WAN, CPU/RAM
  │
MacPro Host (192.168.1.30 · 8-core Xeon E5 · 64GB RAM · macOS 12)
  │  /var/log/**   ←── Promtail
  │  Docker logs   ←── Promtail (docker_sd_configs)
  │  host stats    ←── Node Exporter :9100
  │  custom metrics ←── Telegraf :9273
  │      cpu, mem, disk, processes, SMART, Plex, SMB, security events
  │
  ├── Prometheus   :9090  ← scrapes Node Exporter, Telegraf, Alertmanager
  ├── Loki         :3100  ← receives logs from Promtail
  ├── Alertmanager :9093  ← receives alert rules from Prometheus
  ├── Grafana      :3000  ← queries Prometheus + Loki
  └── cAdvisor     :8080  ← Docker container metrics

---

Quick Start

Prerequisites

• Docker Desktop 4.28.0 (last version supporting macOS 12 Monterey)
  • DMG at /Volumes/4tb-R1/Docker Services/Docker-4.28.0-macOS12.dmg
  • Docker Desktop 4.29+ requires macOS 14 Sonoma

1. Configure credentials

# Set router password for Orbi SOAP scraper
export ROUTER_PASS="your_orbi_admin_password"
# Or store in macOS Keychain:
security add-generic-password -a orbi -s orbi_router -w "your_password"

Edit alertmanager/alertmanager.yml with your email credentials:

smtp_from: 'alerts@yourdomain.com'
smtp_auth_username: 'your@gmail.com'
smtp_auth_password: 'your_gmail_app_password'

2. Start the stack

cd "/Volumes/4tb-R1/Docker Services/NetworkObservability"
./manage.sh start

3. Access Grafana

Open http://localhost:3000 (admin / admin — change on first login)

---

Management Commands

Command	Description
./manage.sh start	Start all services
./manage.sh stop	Stop all services (data preserved)
./manage.sh restart	Restart all services
./manage.sh status	Show container status
./manage.sh logs <service>	Tail logs for specific service
./manage.sh pull	Pull updated images
./manage.sh clean	Remove everything including volumes ⚠️

---

Service URLs

Service	URL	Notes
Grafana	http://localhost:3000	Main dashboard UI
Prometheus	http://localhost:9090	Metrics browser + PromQL
Alertmanager	http://localhost:9093	Alert routing UI
Loki	http://localhost:3100	Log backend (use via Grafana)
Promtail	http://localhost:9080	Log shipper status
Telegraf	http://localhost:9273/metrics	Raw Prometheus metrics
Node Exporter	http://localhost:9100/metrics	Host stats
cAdvisor	http://localhost:8080	Container metrics

---

Grafana Dashboards

All dashboards live in grafana/dashboards/ and are version-controlled.

File	UID	Title	Description
home.json	capes-home	Capes Homelab — Overview	Homepage with links to all dashboards
telegraf-macpro-system.json	telegraf-macpro-system	📊 Telegraf System — Mac Pro	CPU, RAM, disk, SMART, Plex, SMB, security
capes-network-orbi.json	capes-network-snmp	🌐 Network — Orbi Mesh & Capes LAN	Router health, per-device RSSI/speed, latency
node-exporter-full.json	rYdddlPWk	Node Exporter Full	Detailed host metrics (community dashboard)
loki-logs.json	sadlil-loki-apps-dashboard	📋 Log Browser	Loki log explorer

---

Orbi SOAP Scraper (scripts/orbi_metrics.sh)

The Netgear RBR750 does not support SNMP. We use the SOAP API instead.

• Endpoint: https://192.168.1.1/soap/server_sa/
• Auth: HTTP Basic (admin / $ROUTER_PASS)
• Interval: 30 seconds
• Dependency: pure awk/sed/grep — no Python, works in Alpine containers

Metrics produced

Metric	Labels	Description
orbi_up	—	Router reachable (1/0)
orbi_wan_up	wan_ip	WAN connected (1/0)
orbi_cpu_utilization_pct	—	Router CPU %
orbi_memory_utilization_pct	—	Router RAM %
orbi_radio_up	—	WiFi radio status
orbi_radio_info	ssid, channel, security	WiFi info gauge
orbi_firmware_info	firmware, model, serial	Firmware info gauge
orbi_devices_total	—	Total connected devices
orbi_devices_by_connection	conn_type	Count by 2.4GHz/5GHz/wired
orbi_node_device_count	node	Count per mesh node
orbi_device_rssi	name,ip,mac,band,node,type,brand	Per-device signal
orbi_device_linkspeed_mbps	name,ip,band	Per-device link speed
orbi_satellite_up	satellite_mac	Satellite node reachable
orbi_ping_rtt_ms	target,target_ip	Ping RTT to router/Cloudflare/Google
orbi_dns_resolve_ms	—	DNS resolution latency

---

Telegraf Host Metrics

Telegraf runs in Docker, scraping the Mac Pro host via mounted paths and exec scripts.

Inputs configured:

• cpu — per-core and aggregate CPU usage
• mem — RAM used/free/cached/available
• disk — usage % and bytes per volume (/, /Volumes/4tb-R1, /Volumes/6tb-R1, /Volumes/500g-R1)
• processes — running/sleeping/zombie process counts
• [[inputs.exec]] orbi_metrics.sh — Orbi SOAP metrics (see above)
• Custom exec scripts: Plex, SMB, SMART, security events

---

Data Persistence

All data lives in ./data/ (bind-mounted, on 4tb-R1):

Path	Contents	Retention
./data/prometheus/	Metrics time series	90 days
./data/grafana/	Dashboards, users, settings	Indefinite
./data/loki/	Log data	90 days
./data/alertmanager/	Silence records	Indefinite

---

Monitoring the .35 Server

To also monitor 192.168.1.35 (Doris):

1. Install Node Exporter on it (see grafana/dashboards/ for Doris Overview dashboard)
2. Uncomment the node-exporter-server35 job in prometheus/prometheus.yml
3. ./manage.sh restart

---

Migrating to Raspberry Pi

All images have ARM64 builds. To migrate:

1. Copy this entire NetworkObservability/ folder to the Pi
2. Copy ./data/ to preserve historical data
3. curl -sSL https://get.docker.com | sh on the Pi
4. ./manage.sh start

No config changes needed — everything uses container-internal networking.

---

Directory Structure

NetworkObservability/
├── docker-compose.yml
├── manage.sh
├── README.md
├── CHANGELOG.md
├── prometheus/
│   ├── prometheus.yml          # Scrape targets + retention config
│   └── alerts.yml              # Alert rules (disk, CPU, mem, down)
├── loki/
│   └── loki-config.yml
├── promtail/
│   └── promtail-config.yml     # Log sources: /var/log, Docker SD, syslog UDP
├── telegraf/
│   └── telegraf.conf           # Host metrics + Orbi SOAP exec input
├── alertmanager/
│   └── alertmanager.yml        # Alert routing + email (matt@am180.us)
├── scripts/
│   └── orbi_metrics.sh         # Orbi RBR750 SOAP API scraper v2.0.0
├── host-metrics/
│   └── disk_metrics.prom       # Static disk metrics (updated by cron)
├── grafana/
│   ├── provisioning/
│   │   ├── datasources/        # Auto-configured Prometheus + Loki
│   │   └── dashboards/         # Dashboard provisioning config
│   └── dashboards/             # JSON dashboard exports (version-controlled)
└── data/                       # Persistent bind-mount data (not in git)
    ├── prometheus/
    ├── grafana/
    ├── loki/
    └── alertmanager/

---

Pinned Versions

Component	Version
Grafana	10.3.3
Prometheus	2.50.1
Node Exporter	1.7.0
Loki	2.9.4
Promtail	2.9.4
Telegraf	1.29-alpine
Alertmanager	0.26.0
cAdvisor	0.49.1
Docker Desktop	4.28.0 (macOS 12 Monterey max)