iDRAC 9 manual fan control on Dell PowerEdge servers

Kovasky Buezo | Apr 15, 2024 min read

edited on: May 17, 2024

iDRAC 9 manual fan control on Dell PowerEdge servers

Intro

I recently acquired a Dell Poweredge T440, and after installing a GPU the fans skyrocketed. This turned the server into a noise machine. To solve this, I created a script that disables third-party PCI mode and based on a threshold, set the fan speed to 20% or to be controlled by Dell’s fan profile. These are the steps I took to make my T440 silent again.

Downgrading iDRAC 9 firmware to 3.30.30.30

For this whole setup to work, iDRAC needs to be on firmware 3.30.30.30, as later firmware revisions remove the ability to control the fan speed. The firmware can be downloaded from Dell’s website. To downgrade, navigate to your iDRAC’s web admin page, and under Maintenance->System Update you can apply the downgrade. After the downgrade, make sure IPMI over LAN is enabled under iDRAC Settings->Connectivity->Network->IPMI Settings.

Script to control the fan speed

Once iDRAC is on firmware 3.30.30.30 and you can use IPMI over LAN, you can start using my script to control the fan speed. My script first checks that a VM has been up for 60 seconds. I added this check as PfSense is virtualized in my lab and as such needs to wait for it to be up to issue LAN commands. You can remove this check and the script will work as intended. Before running, make sure to fill in the blanks.

#!/bin/bash

check_start_time() {
    local start_time="$1"
    local current_time=$(date +%s)

    local time_diff=$((current_time - start_time))

    if [ $time_diff -ge 60 ]; then
        echo 0
    else
        echo 1
    fi
}

check_temperature() {
    temp_lines=$(ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" sdr type temperature 2>/dev/null | grep -e degrees | awk '{print $(NF-2)}')
    temp_arr=""
    readarray -t temp_arr <<< "$temp_lines"
    sorted_temps=($(printf "%s\n" "${temp_arr[@]}" | sort -nr))
    max_temp=${sorted_temps[0]}

    if [ "$max_temp" -le 60 ] && [ "$CURRENT_PROFILE" != "MANUAL" ]; then
        # enable manual fan control
        ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" raw 0x30 0x30 0x01 0x00 &> /dev/null
        # set initial fan to 20%
        ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" raw 0x30 0x30 0x02 0xff 0x14 &> /dev/null
        LOW_TO_HIGH=false
        CURRENT_PROFILE="MANUAL"
    elif [ "$max_temp" -le 65 ] && [ "$CURRENT_PROFILE" != "MANUAL2" ]; then
        # enable manual fan control
        ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" raw 0x30 0x30 0x01 0x00 &> /dev/null
        # set initial fan to 30%
        ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" raw 0x30 0x30 0x02 0xff 0x1e &> /dev/null
        # set low_to_high used to determine wait time
        if [[ "$CURRENT_PROFILE" == "MANUAL"]]; then 
            LOW_TO_HIGH=true 
        else
            LOW_TO_HIGH=false
        fi 
        CURRENT_PROFILE="MANUAL2"
    elif [ "$max_temp" -gt 65 ] && [ "$CURRENT_PROFILE" != "AUTO" ]; then
        # enable Dell fan control
        ipmitool -I lanplus -H "$HOST" -U "$USER" -P "$PASS" raw 0x30 0x30 0x01 0x01 &> /dev/null
        # set low_to_high used to determine wait time
        if [[ "$CURRENT_PROFILE" == "MANUAL2"]]; then 
            LOW_TO_HIGH=true 
        else
            LOW_TO_HIGH=false
        fi 
        CURRENT_PROFILE="AUTO"
    fi
}

disable_third_party_mode (){
    # disable 3rd party pcie fan mode
    output=$(sshpass -p "$PASS" ssh -o StrictHostKeyChecking=no "$USER@$HOST" "racadm get system.pcieslotlfm" 2>/dev/null)
    pcie_ports_count=$(echo "$output" | grep -c 'System.pcieslotlfm')

    # Iterate through each PCIe port and disable third party mode
    for ((i=1; i<=pcie_ports_count; i++)); do
        sshpass -p "$PASS" ssh -o StrictHostKeyChecking=no "$USER@$HOST" "racadm set system.pcieslotlfm.$i.lfmmode disabled" &> /dev/null
    done
}

HOST=""
USER=""
PASS=""
VMID=""
CURRENT_PROFILE=""
LOW_TO_HIGH=false

start_time=$(ps -o lstart= -p $(cat /var/run/qemu-server/$VMID.pid))
start_time=$(date -d "$start_time" +%s)

while [[ $(check_start_time "$start_time") -eq 1 ]]
do
    sleep 5
done

disable_third_party_mode

while true; do
    check_temperature

    if [[ "$CURRENT_PROFILE" == "AUTO" || "$LOW_TO_HIGH" == true]]; then 
        sleep 180 
    else 
        sleep 60 
    fi
done

Automating the launching of the script

Instead of manually running this every time my server reboots, I created a service. To do so, I used the following service file:

[Unit]
Description=User defined temperature fan control for iDrac 9

[Service]
Type=simple
ExecStart=/usr/local/sbin/fan_control.sh
Restart=always

[Install]
WantedBy=multi-user.target

On my Proxmox install, the bash script is placed at /usr/local/sbin/fan_control.sh with the appropriate permissions and the service file is at /etc/systemd/system. After these files are created, it takes a “systemctl daemon-reload” and then a “systemctl enable fan_control” to start the service and make sure it starts every reboot.

Done!

Once everything is set up, your Dell server with iDRAC 9 should be extremely quiet.