I will explain here how to start and stop boinc depending on the host temperature.
I'm actually using a NUC to run BOINC and the Rosetta@Home project but this host is getting hot very quick and sometimes it will shutdown because of overheating.
To avoid this, I've requested the help of @AenBleidd, one of the BOINC developers. He gave me idea to use lm-sensors and CRON to run a script that will start / stop BOINC.
Based on his idea, I've made the job 😉
To read the host temperature I'll use lm-sensors as adviced by @AenBleidd.
# Install lm-sensors
sudo apt install lm-sensors
# Detect active sensors
sudo sensors-detect
# Run a small test
sensors
# It should return something like that:
coretemp-isa-0000
Adapter: ISA adapter
Package id 0: +42.0°C (high = +100.0°C, crit = +100.0°C)
Core 0: +42.0°C (high = +100.0°C, crit = +100.0°C)
Core 1: +39.0°C (high = +100.0°C, crit = +100.0°C)Unfortunately
lm-sensorsis not working on Raspberry Pi devices so I moved back on theACPIthermal zone information.
Test on Raspberry Pi's:
# Get raw temperature
cat /sys/class/thermal/thermal_zone*/temp
# Get raw rounded temperature value
cat /sys/class/thermal/thermal_zone*/temp | sed 's/\(.\)..$//'
# Get formatted temperature
cat /sys/class/thermal/thermal_zone*/temp | sed 's/\(.\)..$/.\1°C/'The command above are based on this post: https://askubuntu.com/questions/15832/how-do-i-get-the-cpu-temperature/854029#854029
Create a script named boinctempctl and the following content:
#!/bin/bash
# Default config
CONFIG_FILE=/etc/boinctempctl.conf
LOG_FILE=/tmp/boinctempctl.log
LOG_DATE=$(date +"%c")
# Detect Raspberry Pi's
IS_RPI=$(cat /proc/cpuinfo | grep -i bcm2835 > /dev/null ; echo $?)
# Set current temperature based on the device type
if [[ $IS_RPI -eq 0 ]]; then
# Value in celsius degrees
CUR_TEMP=$(cat /sys/class/thermal/thermal_zone*/temp | sed 's/\(.\)..$//')
else
# Value in celsius degrees
CUR_TEMP=$(sensors -u | grep -i temp1_input | cut -d' ' -f4)
fi
# Default limits in celsius degrees
MAX_TEMP=80
MIN_TEMP=40
# Reading config file
[[ -f $CONFIG_FILE ]] && source $CONFIG_FILE
# Check environment
if [ ! -t 1 ]; then
FROM_CRON=true
else
FROM_CRON=false
fi
# Show current temperature
[[ $FROM_CRON == false ]] && echo -e "\nCurrent temperature: ${CUR_TEMP/.000/}°c"
echo -e "\n$LOG_DATE Current temperature: ${CUR_TEMP/.000/}°c" >> $LOG_FILE
# Stop computing if the temp is going too high
if [[ ${CUR_TEMP/.000/} -ge $MAX_TEMP ]]; then
[[ $FROM_CRON == false ]] && echo -e "\nHost is overheating, computing will be stopped.\n"
echo "$LOG_DATE Host is overheating, computing will be stopped." >> $LOG_FILE
if [[ $(systemctl status boinc-client.service | grep -i "active: active" | wc -l) -gt 0 ]]; then
systemctl stop boinc-client.service
#boinccmd --quit
fi
else
if [[ $(systemctl status boinc-client.service | grep -i "active: active" | wc -l) -gt 0 ]]; then
[[ $FROM_CRON == false ]] && echo -e "\nHost temperature is correct, keep computing...\n"
echo "$LOG_DATE Host temperature is correct, keep computing..." >> $LOG_FILE
else
[[ $FROM_CRON == false ]] && echo -e "\nHost is cooling down before restart computing...\n"
echo "$LOG_DATE Host is cooling down before restart computing..." >> $LOG_FILE
fi
fi
# Start computing when the temp is low enough
if [[ ${CUR_TEMP/.000/} -le $MIN_TEMP && ${CUR_TEMP/.000/} -lt $MAX_TEMP ]]; then
[[ $FROM_CRON == false ]] && echo -e "\nHost is fresh enough, start computing.\n"
echo "$LOG_DATE Host is fresh enough, start computing." >> $LOG_FILE
if [[ $(systemctl status boinc-client.service | grep -i "active: active" | wc -l) -eq 0 ]]; then
systemctl start boinc-client.service
fi
fiThen make it executable: chmod -v +x boinctempctl.
The updated script now support configuration file. Change the CONFIG_FILE variable to change the path.
sudo nano /etc/boinctempctl.confAdd the following content:
# boinctempctl config file
# Value in celsius degrees
MAX_TEMP=85
MIN_TEMP=45
# log file
LOG=/tmp/boinctempctl.logYou can change the default values to anything you want.
You can run the script locally but it will require sudo privilege to start / stop the service.
cp -v boinctempctl /home/[user]/sudo cp -v boinctempctl /usr/bin/Now create a CRON task with sudo crontab -e and add the following content:
*/1 * * * * /home/[USER]/boinctempctl
Replace [USER] by your user.
If you have decided to install the script system wide, then change the path to:
/usr/bin/boinctempctl.
And that's all! 🎉
The script will automatically create a log file that you can read with tail -f /tmp/boinctempctl.log.
tail -f /tmp/boinctempctl.log
Wed 22 Apr 2020 01:18:01 AM CEST Host is cooling down before restart computing...
Wed 22 Apr 2020 01:19:01 AM CEST Current temperature: 52°c
Wed 22 Apr 2020 01:19:01 AM CEST Host is cooling down before restart computing...
Wed 22 Apr 2020 01:20:01 AM CEST Current temperature: 48°c
Wed 22 Apr 2020 01:20:01 AM CEST Host is cooling down before restart computing...
Wed 22 Apr 2020 01:21:01 AM CEST Current temperature: 48°c
Wed 22 Apr 2020 01:21:01 AM CEST Host is cooling down before restart computing...You can also simply run the script without CRON. Here are some examples:
# Run the script manually
sudo ./boinctempctl
Wed 22 Apr 2020 01:21:01 AM CEST Current temperature: 48°c
Wed 22 Apr 2020 01:21:01 AM CEST Host is cooling down before restart computing...
# Or like CRON but with watch and below than a minute
watch -n 10 'sudo ./boinctempctl ; tail /tmp/boinctempctl.log'
# Hit [Ctrl + C] to exitIf you have decided to install the script system wide, then change the path to:
/usr/bin/boinctempctl.
Thanks again to @AenBleidd who helped me with his brilliant idea to use lm-sensors and CRON.
Feel free to create a comment to contribute on this gist.
You can reach me on Twitter by using @Jiab77.