English | 繁體中文

Dell iDRAC Fan Controller with GPU Support

A temperature-based fan speed controller for Dell PowerEdge servers (tested on an R730, should work with most PowerEdges). Supports both local and remote hosts.

• Requisites
• Installation / Upgrade
  • Docker
• Configuration
• How it works
• Notes on remote hosts
• Credits

---

Requisites

1. Python 3 is installed.

2. IPMI Over LAN is enabled in all used iDRACs (Login > Network/Security > IPMI Settings).

   • May not be needed if you're only managing the local machine.

3. All hosts to be monitored must have the appropriate sensor tools installed as needed:

   • For monitoring local CPU: install and configure lm-sensors

   • For monitoring NVIDIA GPU: install nvidia-smi

   • For monitoring AMD GPU: install rocm-smi

   • Example output of sensors for a dual CPU system:

     coretemp-isa-0000
     Adapter: ISA adapter
     Core 0:       +38.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 1:       +46.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 2:       +40.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 8:       +43.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 9:       +39.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 10:      +39.0°C  (high = +69.0°C, crit = +79.0°C)
     
     coretemp-isa-0001
     Adapter: ISA adapter
     Core 0:       +29.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 1:       +35.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 2:       +29.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 8:       +34.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 9:       +33.0°C  (high = +69.0°C, crit = +79.0°C)
     Core 10:      +31.0°C  (high = +69.0°C, crit = +79.0°C)
     

Installation / Upgrade

Clone the repo and run the installation script as root to configure the system or upgrade the already installed controller:

git clone https://github.com/kuan909608/dell-idrac-fan-controller-gpu.git
cd dell-idrac-fan-controller-gpu
sudo ./install.sh [<installation path>]

The default installation path is /opt/fan_control and the service will be installed as fan-control.service. If a configuration file already exists, it will be renamed with a .old extension.

Docker

To deploy remote fan management with Docker (fan_control running on a separate host and only interacting with remote ones, see Notes on remote hosts), build the image in the repo and bind mount your own YAML config and SSH keys folder:

git clone https://github.com/kuan909608/dell-idrac-fan-controller-gpu.git
cd dell-idrac-fan-controller-gpu
docker build -t fan_control .
docker run -d --restart=always --name fan_control -v "./fan_control_config.yaml:/app/fan_control_config.yaml:ro" -v "./keys:/app/keys:ro" fan_control

Running this tool under a proper orchestrator is advised.

---

Deployment Method Selection Guide

You can deploy this tool in two ways: systemd (bare-metal) or Docker. Please choose only one method for each host.

When to use systemd (bare-metal)

• Recommended if you need direct access to hardware sensors (e.g., lm-sensors) on the host.
• Suitable for environments where you want the service to start automatically with the OS and be managed by systemd.
• The install.sh script automates dependency installation, venv setup, file copying, and systemd service configuration.

When to use Docker

• Recommended for remote-only management, or if you want to isolate the environment and simplify migration.
• If you need to access hardware sensors inside Docker, you must mount additional system directories (e.g., /dev, /sys). Example:

  docker run ... -v /dev:/dev -v /sys:/sys ...

• Make sure to mount your configuration file and SSH keys as shown above.
• For production, use an orchestrator for better reliability.

Important Notes

• Do not enable both systemd service and Docker container on the same host at the same time. Running both may cause conflicts or resource contention.
• The install.sh script will overwrite existing files and systemd service. Backup your configuration before running it.
• When using SSH keys in Docker, ensure proper permissions and security practices.

Configuration

You can tune the controller's settings via the fan_control_config.yaml file in the installation directory.

Configuration File Structure

The configuration file contains two main sections: general and hosts.

general section

Key	Description
debug	Toggle debug mode (print ipmitool commands instead of executing them, enable additional logging).
interval	How often (in seconds) to read the CPUs' and GPUs' temperatures and adjust the fans' speeds.
temperature_control_mode	Use max or avg to decide if fan control is based on the maximum or average temperature.
cpu_temperature_command	Shell command to get CPU temperatures (semicolon separated).
gpu_temperature_command_nvidia	Shell command to get NVIDIA GPU temperatures (semicolon separated).
gpu_temperature_command_amd	Shell command to get AMD GPU temperatures (semicolon separated).

hosts section

Each host object supports the following keys:

Key	Description
name	Host name identifier.
fan_control_mode	Fan control mode, manual or automatic.
temperatures	List of temperature thresholds (in °C). Must have at least 2 values.
speeds	List of fan speeds (in %) for each threshold. Must have at least 2 values.
hysteresis	Hysteresis value in °C to prevent rapid fan speed changes.
ipmi_credentials	(Optional) IPMI login info for this host.
ssh_credentials	(Optional) SSH login info for this host. Supports host, username, password, and optional key_path for SSH key authentication.
gpu_type	(Optional) Supported GPU types, can be a string (e.g., nvidia) or an array (e.g., [nvidia, amd]).
vms	(Optional) List of VM objects. See below for VM object structure.

vms objects

Each VM object supports the following keys:

Key	Description
name	VM name identifier.
ssh_credentials	SSH login info for the VM. Supports host, username, password, and optional key_path for SSH key authentication.
gpu_type	Supported GPU types for the VM, can be a string (e.g., nvidia) or an array (e.g., [nvidia, amd]).

Auto-splitting thresholds and speeds

If you only specify 2 pairs of temperatures and speeds (e.g., [40, 80] and [20, 80]), the system will automatically split them into multiple steps based on the hysteresis value.

The splitting logic:

• The range between temp_min and temp_max will be divided into intervals of hysteresis * 2.
• For each interval, a new threshold and corresponding speed will be generated, resulting in smoother fan speed transitions.

Example:

temperatures: [40, 80]
speeds: [20, 80]
hysteresis: 5

This will be automatically expanded to:

thresholds: [40.00, 50.00, 60.00, 70.00, 80.00]
speeds: [20, 35, 50, 65, 80]

Example

general:
  debug: False
  interval: 60
  cpu_temperature_command: "sensors | grep -E 'Core [0-9]+:' | awk '{print $3}' | sed 's/+//;s/°C//' | paste -sd ';' -"
  gpu_temperature_command_nvidia: "nvidia-smi --query-gpu=temperature.gpu --format=csv,noheader,nounits | paste -sd ';' -"
  gpu_temperature_command_amd: "rocm-smi --showtemp | grep -E 'Temp' | awk '{print \$2}' | sed 's/[^0-9.]//g' | paste -sd ';' -"

hosts:
  - name: host1
    temperatures: [40, 60, 80]
    speeds: [20, 50, 80]
    hysteresis: 5
    ipmi_credentials:
      host: 10.0.0.1
      username: admin
      password: password
    ssh_credentials:
      host: 10.0.0.2
      username: admin
      password: password
    gpu_type: nvidia
    vms:
      - name: vm1
        ssh_credentials:
          host: 10.0.0.3
          username: user
          password: password
        gpu_type: [nvidia]
  - name: host2
    temperatures: [35, 55, 75]
    speeds: [30, 60, 90]
    hysteresis: 5
    gpu_type: nvidia

Auto-splitting thresholds and speeds

If you only specify 2 pairs of temperatures and speeds (e.g., [40, 80] and [20, 80]), the system will automatically split them into multiple steps based on the hysteresis value.

The splitting logic:

• The range between temp_min and temp_max will be divided into intervals of hysteresis * 2.
• For each interval, a new threshold and corresponding speed will be generated, resulting in smoother fan speed transitions.

Example:

temperatures: [40, 80]
speeds: [20, 80]
hysteresis: 5

This will be automatically expanded to:

thresholds: [40.00, 50.00, 60.00, 70.00, 80.00]
speeds: [20, 35, 50, 65, 80]

Additional Notes

• temperature_control_mode (in general): Set to max to use the highest temperature for fan control, or avg to use the average.
• fan_control_mode (in each host): Set to manual for script-controlled fan speed, or automatic to let hardware manage it.
• gpu_type: Can be a string or an array, e.g., nvidia or [nvidia, amd].
• ssh_credentials.key_path: (Optional) Path to SSH private key for authentication.
• VM objects under vms also support gpu_type as an array.

Multi-host and VM Support

• Each host can define its own temperature thresholds, fan speeds, and credentials.
• Hosts can include a vms list，each VM supports its own SSH credentials and GPU type。
• The script will collect GPU temperatures from both the host and all defined VMs, and use the highest temperature for fan control.

Temperature Collection & Fan Control Logic

• The script collects CPU and GPU temperatures for each host and its VMs.
• If temperature data is missing or abnormal, the fan will run at the highest configured speed (the last value in speeds) for safety.
• The maximum value among all CPU/GPU temperatures (including VMs) is used as the control temperature.
• Fan speed is set according to the configured thresholds and speeds.
• All temperature readings and control actions are recorded in the internal state for monitoring and debugging.

How it works

Every interval seconds, the controller will collect CPU and GPU temperatures from all hosts and their VMs.
The highest temperature among all CPUs/GPUs (including VMs) is used as the control temperature to determine the fan speed.

• If temperature data is missing or abnormal, the fan will run at the highest configured speed (the last value in speeds) for safety.
• Fan speed is set according to the configured thresholds and speeds.
• All temperature readings and control actions are recorded in the internal state for monitoring and debugging.

Fan speed is determined by each temperature threshold and its corresponding speed. The number of thresholds/speeds can be any matching pair count.

Condition	Fan speed
Tmax ≤ Threshold1	Speed1
Threshold1 < Tmax ≤ Threshold2	Speed2
...	...
Tmax > ThresholdN	Highest configured speed (last value in speeds)

If hysteresis is set for a given host, the controller will wait for the temperature to go below ThresholdN - hysteresis before lowering the fan speed.
For example: with a Threshold2 of 37°C and a hysteresis of 3°C, the fans won't slow down from Threshold3 to Threshold2 speed until the temperature reaches 34°C.

Notes on remote hosts

This controller can monitor the temperature and change the fan speed of remote hosts too: the only caveat is that you'll need to extract the temperatures via an external command. This could be via SSH, for example. The controller expects such a command to return a newline-delimited list of numbers parseable as floats.

The included example is a good fit for a remote Proxmox VE host: it will connect to it via SSH and extract the temperature of all CPU cores, one per line. This way you'll be able to manage that machine just as well as the local one without applying any hardly trackable modification to the base OS.

Credits

Major thanks go to NoLooseEnds's directions for the core commands, sulaweyo's ruby script for the idea of automating them, and nmaggioni's r710-fan-controller as the main forked project.

Note: The key difference of this script, other than handling remote hosts, is that it's based on the temperature of the CPUs' cores and not on the ambient temperature sensor on the server's motherboard.