I've created a bash script that collects data from a solar inverter every 10 seconds for a duration of 5 minutes. It performs some calculations and then sends the data to a platform. While it works, I'm noticing that the CPU usage is quite high, peaking at about 80%. I'm looking for an effective way to manage the timing for the next data collection without running a time-consuming loop. Specifically, I want to improve line 7 of my script. Here's a simplified version of it:
```bash
#!/bin/bash
set -o pipefail
IFS=$''
samples="0"
nr="0"
while [ $SECONDS -lt 292 ]; do #5min-8s
if [[ (( $(( (samples - 1) * 10 + 10 )) == $SECONDS )) || (( 0 == $SECONDS )) ]]; then
((samples++))
timestart=$SECONDS
output="$(./inverter_poller --run-once)" # get data from inverter
timeend=$SECONDS
echo ${output} > /var/log/inverter.last
rs232time=$((timeend - timestart)) # usually it is 6-7 seconds
if (( rs232time > /var/log/inverter.last
fi
looptime=$((SECONDS - timestart))
echo "time": $looptime >> /var/log/inverter.last
fi
done
***boring data processing and sending to emoncms was here***
```
Any suggestions for optimizing my script?
1 Answer
Your issue seems to stem from the continuous looping of your script, resulting in high CPU usage. Instead, you might want to introduce a sleep command after you fetch the data. For instance:
```bash
while [ $SECONDS -lt 292 ]; do
((samples++))
timestart=$SECONDS
output="$(./inverter_poller --run-once)" # get data from inverter
timeend=$SECONDS
echo ${output} > /var/log/inverter.last
rs232time=$((timeend - timestart)) # usually 6-7 seconds
if (( rs232time > /var/log/inverter.last
fi
looptime=$((SECONDS - timestart))
echo "time": $looptime >> /var/log/inverter.last
sleep 10
done
```
This will effectively create a 10-second pause between each data collection and should help reduce CPU spikes significantly. You can even adjust the sleep duration depending on your specific timing needs.
That's a good strategy! Also, if your rs232time consistently exceeds 10 seconds, you might want to log that and alert yourself instead of losing the readings.
I ran into issues with sleep because my loop was often taking longer than expected on some iterations. One workaround I found is to check the time taken and only sleep for the remaining duration:
```bash
if (( $looptime < 10 )); then sleep $((10 - $looptime)); fi ``` Just ensure that you handle cases where the operation might take longer than 10 seconds.