Proxmox CPU Temperature Monitor for InfluxDB

You notice the fans ramping on a Proxmox box, open the web UI, and get the usual helpful answer: almost nothing about actual CPU temperature. Great. Very enterprise. Very "go SSH into the host like it is 2009."

This project is a small shell-based Proxmox CPU temperature monitor. It reads CPU sensor data on the host, formats the readings as metrics, and sends them to InfluxDB so the numbers can be graphed, compared, and alerted on instead of trapped in a one-off terminal session.

What problem does it solve?

Proxmox VE is excellent at managing VMs and containers, but hardware temperature visibility depends on the host, sensors, and whatever monitoring stack you bolt on. If you want to answer "what is my Proxmox CPU temperature right now?" or "did the server get hotter after I moved that workload?", you need a repeatable collector.

This monitor keeps the boring part boring:

How it works

The script runs on the Proxmox host, reads CPU temperature values from Linux sensor tooling, and writes them into InfluxDB. The intended setup is deliberately simple: shell script, host cron or service runner, and an InfluxDB bucket/database.

Typical flow:

1. Install and verify lm-sensors on the Proxmox host.
2. Configure the InfluxDB endpoint, token or credentials, org, and bucket/database.
3. Run the collector manually once and confirm metrics are accepted.
4. Schedule it with cron or a systemd timer.
5. Build a dashboard or alert from the stored temperature series.

Example setup

Install sensor support on the Proxmox host:

apt update
apt install lm-sensors
sensors-detect
sensors

Run the monitor with your InfluxDB configuration:

export INFLUX_URL="http://influxdb.local:8086"
export INFLUX_TOKEN="your-token"
export INFLUX_ORG="homelab"
export INFLUX_BUCKET="proxmox"

./proxmox-cpu-temp-monitor.sh

For unattended monitoring, schedule the collector every minute:

* * * * * /opt/proxmox-cpu-temp-monitor/proxmox-cpu-temp-monitor.sh >> /var/log/proxmox-cpu-temp-monitor.log 2>&1

Example output

A truthful monitoring page needs actual output, not a heroic stock image of a server looking moody. The collector produces temperature metrics that can be stored and graphed over time:

proxmox_cpu_temperature,host=pve-node-01,chip=k10temp,core=tctl value=52.8
proxmox_cpu_temperature,host=pve-node-01,chip=k10temp,core=ccd1 value=48.5
proxmox_cpu_temperature,host=pve-node-01,chip=coretemp,core=package_id_0 value=55.0

That gives you a clean time series for questions like:

• Is this Proxmox node idling hotter than usual?
• Did a VM migration or backup job create a thermal spike?
• Is a fan curve, dust filter, thermal paste job, or case airflow change actually helping?

Dashboard ideas

The most useful dashboard is not fancy. It just needs to make the bad number obvious before it becomes a worse number.

Where it fits

This belongs next to the other tiny infrastructure tools I use to remove recurring papercuts. Portscope catches Docker port conflicts before containers launch. PZEM ESPHome Integration tracks electrical data in Home Assistant. This one handles the thermal side of a Proxmox homelab.

Different dashboards. Same principle: make invisible machine state visible before it bites you.

When to use it

Use this if:

• You run Proxmox VE on a homelab, mini PC, workstation, or small server.
• You already use InfluxDB or Grafana for host metrics.
• You want a lightweight CPU temperature collector instead of a full monitoring platform.
• You care about heat trends during backups, VM migrations, AI workloads, or summer afternoons that turn your rack into a toaster.

Do not use it if you need a full observability stack with service discovery, distributed tracing, and a dashboard that requires three meetings to interpret. That is a different genre of suffering.

Source

The project is open source on GitHub. The page canonical is https://bnap.dev/projects/proxmox-cpu-temp-monitor.