lockerstor as6604t fan control script - beta version

temp_monitor6604t.sh

#!/bin/bash

# quick and dirty fan control script
#
# updated version that handles different hardware dynamically
#
# v2.1 30-8-2022 ( initial test version )
# v2.2 31-8-2022 ( abstracted hwmon paths and core/drive counts - should now handle any intel cpu and number of drives )
#
#
# depends on:
#
# lm_sensors (to query system sensors)
#
# smartmontools + hddtemp (to query hdd temp sensors)
#
# custom it87 kmod to read/set fan speeds
#    needs to be compiled from source ( https://github.com/a1wong/it87 ) as the debian supplied
#    it87 doesn't support the IT8625E used in the asustor



# uses sqr(temp above threshold)+base_pwm for hdd response curve
# uses sqr(temp above threshold/2)+base_pwm for sys response curve
#
# this gives a slow initial ramp up and a rapid final ramp up across the desired temp range (hdd range 35-50 celsius sys range 50-75 celsius )



#
# global variables to tune behaviour
#

# debug output level : 0 = disabled 1 = minimal 2 = verbose 3 = extremely verbose
debug=0

# enable email fan change alets
mailalerts=0

# how often we check temps / set speed ( in seconds )
frequency=10

# ratio of how often we update system sensors vs hdd sensors 
# sampling the sys sensors is lightweight, wheras querying the hdd sensors via SMART disrupts disk i/o - and hdd temp doesn't change that fast
ratio_sys_to_hdd=12

# the hdd temperature above which we start to increase fan speed
hdd_threshold=35

# the system temperatures above which we start to increase fan speed
sys_threshold=50

# minimum pwm value we ever want to set the fan to ( 70 == 1600 rpm, 60 == 1400 )
min_pwm=60


#
# determine the /sys/class/hwmon mappings
#

# which /sys/class/hwmon symlink points to the it87 ( fan speed )
hwmon_it87="/sys/class/hwmon/"`ls -lQ /sys/class/hwmon | grep -i it87 | cut -d "\"" -f 2`
if [ $debug -gt 1 ]; then
   echo "hwmon_it87 is "  $hwmon_it87
fi

# which /sys/class/hwmon symlink points to the intel coretemp sensors
hwmon_coretemp="/sys/class/hwmon/"`ls -lQ /sys/class/hwmon | grep -i coretemp | cut -d "\"" -f 2`
if [ $debug -gt 1 ]; then
   echo "hwmon_coretemp is " $hwmon_coretemp
fi

# which /sys/class/hwmon symlink points to the acpi sensors ( board temperature sensor )
hwmon_acpi="/sys/class/hwmon/"`ls -lQ /sys/class/hwmon | grep -i thermal_zone0 | cut -d "\"" -f 2`
if [ $debug -gt 1 ]; then
   echo "hwmon_acpi is " $hwmon_acpi
fi




# set fan speed to desired_pwm
function set_fan_speed() {

    # with the it87 module loaded fan speed is readable via /sys/class/hwmon/hwmonX - the fan speed is on pwm1
    # 0 == full speed 255 = stopped
    # note that this is inverted compared to ADM fanctrl which has 0 = stopped 255 = max 

    local real_pwm
    let real_pwm=255-$desired_pwm
    echo $real_pwm >$hwmon_it87/pwm1

}



# query fan speed and set the global fan_rpm
function get_fan_speed() {

    fan_rpm=`cat $hwmon_it87/fan1_input`

}


# query all drive temperatures and set the global hdd_temp to the highest
function get_hdd_temp() {

    # presume we have up to 8 drives - find the highest temperature on them
    hdd_temp=`hddtemp -n /dev/sd[a,b,c,d,e,f,g] | sort -rn | head -1`

    if [ $debug -gt 1 ]; then
       echo "hdd_temp: " $hdd_temp
    fi

}




# query system temperatures and set the global sys_temp with the highest
function get_sys_temp() {

    # read the system board temp sensor via acpi
    local acpi_temp=`cat $hwmon_acpi/temp1_input`
    acpi_temp=$(expr $acpi_temp / 1000)

    # read all the temps available via coretemp ( pkg + core1..N ) and return the highest
    local cpu_temp=`cat $hwmon_coretemp/temp?_input | sort -nr | head -1`
    cpu_temp=$(expr $cpu_temp / 1000)

    # choose the greatest of the core and system temps
    sys_temp=$(( $acpi_temp > $cpu_temp ? $acpi_temp : $cpu_temp ))

    if [[ $debug -gt 1 ]] ; then
       echo "acpi temp:" $acpi_temp " cpu_temp:" $cpu_temp " sys_temp: " $sys_temp
    fi

}




# map the current hdd_temp to a desired pwm value
#
# we use base_pwm_value+sqr(hdd_temp-hdd_threshold) to get a nice curve
#
function map_hdd_temp() {
     if [[ $hdd_temp -le $hdd_threshold ]] ; then

          if [[ $debug -gt 1 ]]; then
             echo "hdd temp (" $hdd_temp ") is under threshold"
          fi

          hdd_desired_pwm=$min_pwm

     else

          if [[ $debug -gt 1 ]] ; then
             echo "hdd temp (" $hdd_temp ") is over threshold"
          fi

          # get the difference above threshold
          let hdd_desired_pwm=$hdd_temp-$hdd_threshold
          # square it
          let hdd_desired_pwm=$hdd_desired_pwm*$hdd_desired_pwm
          # add it to the base_pwm value
          let hdd_desired_pwm=$min_pwm+$hdd_desired_pwm

     fi

     if [[ $hdd_desired_pwm -gt 255 ]] ; then
          # over max - truncate to max
          hdd_desired_pwm=255
     fi

     if [[ $debug -gt 2 ]] ; then
        echo "+++ map_hdd_temp - hdd_desired_pwm : " $hdd_desired_pwm
     fi
}




# map the current sys_temp to a desired pwm value
#
# we use base_pwm_value+sqr((hdd_temp-hdd_threshold)/2) to get a nice curve
#
function map_sys_temp() {
     if [[ $sys_temp -le $sys_threshold ]] ; then

          if [[ $debug -gt 1 ]] ; then
             echo "sys temp (" $sys_temp ") is under threshold"
          fi

          sys_desired_pwm=$min_pwm

     else

          if [[ $debug -gt 1 ]] ; then
             echo "sys temp (" $sys_temp ") is over threshold"
          fi

          # get the difference above threshold
          let sys_desired_pwm=$sys_temp-$sys_threshold
          # halve the difference
          let sys_desired_pwm=$sys_desired_pwm/2
          # then square it
          let sys_desired_pwm=$sys_desired_pwm*$sys_desired_pwm
          # add it to the base pwm value
          let sys_desired_pwm=$min_pwm+$sys_desired_pwm

     fi

     if [[ $sys_desired_pwm -gt 255 ]] ; then
          # over max - truncate to max
          hdd_desired_pwm=255
     fi

     if [[ $debug -gt 2 ]] ; then
        echo "+++ map_sys_temp - sys_desired_temp : " $sys_desired_pwm
     fi
}




# determine desired zone based on current temp
function get_desired_pwm() {

    map_sys_temp
    map_hdd_temp

    if [[ $hdd_desired_pwm -gt $sys_desired_pwm ]] ; then
         desired_pwm=$hdd_desired_pwm
         if [[ $debug -gt 2 ]] ; then
            echo ">>> choosing hdd pwm value - desired pwm " $desired_pwm
         fi
    else
         desired_pwm=$sys_desired_pwm
         if [[ $debug -gt 2 ]] ; then
            echo ">>> choosing sys pwm value - desired pwm " $desired_pwm
         fi
    fi
}

## MAIN #################################################################################

# get initial temperatures

get_sys_temp
get_hdd_temp
get_fan_speed

last_sys_temp=$sys_temp
last_hdd_temp=$hdd_temp

# we use the variables 'last_pwm' 'last_sys_temp' and 'last_hdd_temp' to track what the pwm/temps values were last time 
# thru the loop - so we only change the fan speeds when there's a state change  as opposed to every iteration

# get initial pwm value

get_desired_pwm

last_pwm=$desired_pwm

# set initial fan speed

if [[ $debug -gt 1 ]] ; then
   echo "initial fan pwm " $desired_pwm
fi

set_fan_speed

# now loop forever monitoring and reacting

cycles=$ratio_sys_to_hdd

while true; do

    # update sensor readings

    get_fan_speed

    get_sys_temp

    if [[ $cycles -eq 1 ]] ; then

       if [[ $debug -gt 1 ]] ; then
          echo "sampling hdd sensor"
       fi

       get_hdd_temp
       cycles=$ratio_sys_to_hdd

    else

       if [[ $debug -gt 1 ]] ; then
          echo "skipping hdd sensor update"
       fi

       let cycles=$cycles-1

    fi

    # update target pwm value based on readings

    get_desired_pwm

    if [[ $debug -gt 1 ]] ; then
         echo "desired pwm " $desired_pwm " last_pwm " $last_pwm
         echo "sys_temp " $sys_temp " last_sys_temp " $last_sys_temp
         echo "hdd_temp " $hdd_temp " last_hdd_temp " $last_hdd_temp
    fi

    if [[ $desired_pwm -gt $last_pwm ]] ; then
       # fan speed increase desired - react immediately

       if [[ $debug -ge 1 ]] ; then
          echo "!!!! fan speed INCREASE : hdd_temp " $hdd_temp " sys_temp " $sys_temp " fan_rpm " $fan_rpm " - changing fan speed to " $desired_pwm
       fi

       if [[ $mailalerts -ge 1 ]] ; then
          echo "!!!! fan speed INCREASE : hdd_temp " $hdd_temp " sys_temp " $sys_temp " fan_rpm " $fan_rpm " - changing fan speed to" $desired_pwm | mail root@as5202t.home -s "AS5202T - temperature alert"
       fi

       # set the fan speed

       set_fan_speed

       # update state tracking variables ONLY when there's a change in the target fan speed

       last_pwm=$desired_pwm
       last_sys_temp=$sys_temp
       last_hdd_temp=$hdd_temp
    fi

    if [[ $desired_pwm -lt $last_pwm ]] ; then
       # fan speed decrease desired
 
       # calculate deltas from last reading for each sensor

       let hdd_delta=$last_hdd_temp-$hdd_temp
       let sys_delta=$last_sys_temp-$sys_temp

       if [[ $debug -gt 1 ]] ; then
          echo "current sys_delta " $sys_delta " current hdd_delta " $hdd_delta
       fi

       # we need to apply 2 degrees of hysteresis on hdd_temp and 4 degrees on sys_temp to prevent fan speed hunting

       if [[ $hdd_delta -gt 2 ]] || [[ $sys_delta -gt 4 ]] ; then   

          # we've got sufficient downward temp delta - actually change the fan speed

          if [[ $debug -ge 1 ]] ; then
             echo "!!!! fan speed DECREASE : hdd_temp " $hdd_temp " sys_temp " $sys_temp " fan_rpm " $fan_rpm " - changing fan speed to " $desired_pwm
          fi
          if [[ $mailalerts -ge 1 ]] ; then
             echo "!!!! fan speed DECREASE : hdd_temp " $hdd_temp " sys_temp " $sys_temp " fan_rpm " $fan_rpm " - changing fan speed to" $desired_pwm | mail root@as5202t.home -s "AS5202T - temperature alert"
          fi

          # set the fan speed

          set_fan_speed

          # update state tracking variables ONLY when there's a change in the target fan speed

          last_pwm=$desired_pwm
          last_sys_temp=$sys_temp
          last_hdd_temp=$hdd_temp

       else

          # not enough downward delta to trigger an actual change yet

          if [[ $debug -ge 1 ]] ; then
             echo "!!!! fan speed DECREASE : hdd_temp " $hdd_temp " sys_temp " $sys_temp " - but not enough delta (" $hdd_delta " " $sys_delta ") yet!"
          fi
       fi
    fi

    sleep $frequency

done

if you don't want email alerts change the mailalerts variable ( line 33 ) to 0 - similarly once you're sure the temperature monitoring and fan response is working you can change debug ( line 30 ) to 0 so you don't get piles of output cluttering up the console/system logs

I wanted to Thank You for modifying your script for my box, I wanted to see if it was ok if I posted on the TrueNAS Forums to share with others that might be using the same box as me. I don't want to post without your permission.

absolutely no problem - I'll go link the '6604t' version from the main gist though as the background stuff on installng lm_sensors etc is in there, so just share the main ( public ) link https://gist.github.com/johndavisnz/bae122274fc6f0e006fdf0bc92fe6237

I've now got a alpha version running which dynamically determines the hwmon mappings and deals with variable numbers of drives and cores, which means it should work for pretty much any intel based NAS model - but I'll do some more testing on it before posting it

Nice I’m sure that will help many people that had the ransomware attack this year and want to get away from their NAS ADM like me. Amazing work! From: johndavisnz ***@***.***> Sent: Tuesday, August 30, 2022 12:32 AM To: johndavisnz ***@***.***> Cc: OrcD3vil ***@***.***>; Comment ***@***.***> Subject: Re: johndavisnz/temp_monitor6604t.sh @johndavisnz commented on this gist.

…

________________________________ I've now got a alpha version running which dynamically determines the hwmon mappings and deals with variable numbers of drives and cores, which means it should work for pretty much any intel based NAS model - but I'll do some more testing on it before posting it — Reply to this email directly, view it on GitHub<https://gist.github.com/6f21205fa77414bfccd8a44b43c72afb#gistcomment-4283968>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A2ZWX6YQGFRYV7JTHWQ477DV3WMF3ANCNFSM577U75IQ>. You are receiving this because you commented.Message ID: ***@***.******@***.***>>

system independent version works on the as5202t - if you can confirm it works on the 6604t I'll update the main gist page with this version

It is working for me I changed the debug and mail and it seems to ramp up and down based on heat. Thank you so much! Only thing is to finish setting it up and see if someone figured out the LCD display hehe

on earlier models the LCD is on a serial port - I'd start here and see what works : https://github.com/mafredri/lcm

I've updated the version on the main page and removed the link to this beta version ( any future updates will be to the main version )