Rack Power Monitor

Per-rack power monitoring for APC PDUs. Combines kW draw across multiple PDUs via SNMP and alerts before you hit your colo power cap.

Features

• SNMP polling of APC AP8841 PDUs (rPDU2 MIB, SNMP v2c)
• Per-rack aggregation — sums power from configured PDUs per rack
• Web dashboard with OK / Warning / Danger states and per-PDU breakdown
• Configuration UI for PDUs, thresholds, poll interval, and SMTP alerts
• Email alerts with cooldown (warning, critical, PDU unreachable, recovery)
• Webhook alerts for Slack, Discord, or custom JSON endpoints
• 24-hour power history sparkline per rack on the dashboard
• SNMP v3 support (auth/priv protocols configurable)
• Maintenance mode to silence alerts during planned work
• Config export/import (download/upload YAML from the config page)
• Stale/unreachable PDUs are clearly flagged — never silently shown as zero
• Single YAML config file — no hardcoded IPs or credentials

Quick start

1. Copy and edit config

cp config.example.yaml config.yaml
# Edit config.yaml with your PDU IPs, community strings, racks, and SMTP settings

2. Verify SNMP OIDs (important)

APC firmware returns power in scaled integer units. Confirm with a live walk:

snmpwalk -v2c -c public 10.0.1.11 1.3.6.1.4.1.318.1.1.26.4.3.1.5

Adjust snmp.power_divisor in config until the dashboard matches your expected kW reading. Common values:

Raw SNMP value	Divisor	Result
1410 (watts)	1000	1.41 kW
14100 (tenths of W)	10000	1.41 kW

3. Run locally

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
rack-power-monitor
# or: python -m rack_power_monitor.main

Open http://localhost:8080 for the dashboard and http://localhost:8080/config for settings.

Mac users: see MAC_SETUP.md for full step-by-step instructions.

4. Docker

cp config.example.yaml config.yaml
# edit config.yaml
docker build -t rack-power-monitor .
docker run -d -p 8080:8080 -v $(pwd)/config.yaml:/app/config.yaml rack-power-monitor

Configuration

All settings live in config.yaml:

Section	Purpose
poll_interval_seconds	How often to poll PDUs (30–60s typical)
alert_cooldown_minutes	Minimum time between repeat alert emails
snmp	OIDs, timeout, power scaling divisor
racks	Rack name, description, thresholds, PDU list
smtp	Host, port, TLS, auth, recipients
webhooks	Slack/Discord/custom webhook URLs and format

Threshold logic (kW draw)

Per rack, combined instantaneous power from all reachable PDUs:

• OK (green): below warning threshold (default 2.5 kW)
• Warning (yellow): at or above warning threshold
• Danger (red): at or above critical threshold (default 3.0 kW)

Note: Colo caps are almost always kW (instantaneous draw), not kWh (energy over time). Rack Power Monitor tracks kW draw. Energy OID is polled for display but not used for threshold alerts in v1.

API endpoints

Method	Path	Description
GET	/api/status	Current rack/PDU readings
POST	/api/refresh	Force immediate poll
GET/PUT	/api/config	Read/save configuration
POST	/api/test/pdu	Test single PDU SNMP
POST	/api/test/pdu/all	Test all configured PDUs
POST	/api/test/smtp	Send test alert email
POST	/api/test/webhooks	Send test webhook
GET	/api/config/export	Download config as YAML
POST	/api/config/import	Upload and apply YAML config

Project structure

rack_power_monitor/   Python package (SNMP, polling, alerts, API)
templates/          Dashboard and config HTML
static/             CSS and JavaScript
config.example.yaml Example configuration
Dockerfile          Container build

License

MIT