Re: Raspberry Pi b und Systemlast
Posted: Wed Dec 24, 2014 1:23 pm
looking into this I realized that on your system the job that computes "every minute" data in "/home/meteohub/histeval0" takes about 30 seconds to run when you enter the end of the month. When other computations are also running, then it might take longer than a minute and by that histeval0 calls (which are done every minute) do stack up.
I made an experimental change in your histeval0 file by checking that new computation is only started when no old one still being done. This might result into updating the minute data sometimes a minute later, but as the data for the last hour is computed, a missing minute will be closed on the next run.
Lets see, if this fixes the issue. I will put the code then into production.
As you see, there has been another technique reducing computation calls when using the slow NSLU2, but to check if a job is still running seems to be the better way.
I made an experimental change in your histeval0 file by checking that new computation is only started when no old one still being done. This might result into updating the minute data sometimes a minute later, but as the data for the last hour is computed, a missing minute will be closed on the next run.
Lets see, if this fixes the issue. I will put the code then into production.
As you see, there has been another technique reducing computation calls when using the slow NSLU2, but to check if a job is still running seems to be the better way.
Code: Select all
#!/bin/sh
if [ "$METEOHUBHW" == "NSLU2" ]
then
mod=$(((`date +"%s"` / 60) % 3))
if [ $mod -gt 0 ]
then
exit 0
fi
fi
if [ `ps ax | grep -e "-c min1 -t -3720" | grep -v grep | wc -l` -lt 1 ]
then
/home/meteohub/wmr928eval -c min1 -t -3720 -w /data/weather/ -s /home/meteohub/wmr928eval.conf -S
fi