I just found out rtl_433 is the cause of the 70% nonstop cpu usage. I 
optimized the rtl_433 command:

cmd = rtl_433 -f 868.3M -M time:utc -M protocol -M level -F json -Y classic 
-Y autolevel -Y squelch -s 250k -R 78 -R 32
(added squelch, autolevel and reduced bandwith to 250k).

CPU usage down to 15%, let's see how this goes. Maybe I will try out 
whether neowx-material will still occupy the cpu for so long.


[email protected] schrieb am Freitag, 4. März 2022 um 10:57:09 UTC+1:

> Or could I send the live data from one raspberry to another (or any other 
> machine with more performance reserves) to another weewx installation with 
> own database, which then does the reports for the heavier skins? Something 
> like a realtime.txt or IP port from one weewx machine to another? I know I 
> can do it partially with FOSHKplugin for ecowitt-client data, but how could 
> i forward weewx data from SDR and USB?
>
> [email protected] schrieb am Freitag, 4. März 2022 um 10:13:32 UTC+1:
>
>> With this skin enabled, cpu usage stays for around 50 out of 60 seconds 
>> at 100%. It also skips some reports as the old one isnt finished (found in 
>> log).
>>
>> If i disable it (removing from weewx conf) cpu mostly stays at 70% and 
>> peaks to 100 for around 10 to 20 seconds. RAM usage is around 350 MB. 
>>
>> Is there a way to only let it run let's say every 10 minutes, while 
>> letting the seasons skin run every minute? Skin dependent reporting 
>> interval possible?
>>
>>
>> [email protected] schrieb am Freitag, 4. März 2022 um 08:38:11 UTC+1:
>>
>>> How many weewx-instances and skins can be run simultaneously with 
>>> acceptable performance on a Raspberry Pi 3B and 4?
>>>
>>> I wanted to put as many weewx instances as possible on one RPi3 B as I 
>>> have an usv/ups board (StromPi 2) installed on it. Multiple RPi3/4 with a 
>>> battery backup each would get a bit expensive.
>>>
>>> In its final form it should run at least 4 instances, maybe with the 
>>> possibility to add one more sensor in the future:
>>>
>>> 2x SDR instances for 2 different outdoor sensor arrays a / b
>>> 1x interceptor driver with FOSHKplugin for sensor array a
>>> 1x USB driver for sensor array b which is recieved with the USB-console
>>>
>>> The both SDR instances will also record some sensors of the secondary 
>>> sensor array each, so:
>>> sensor array a primary + sensor array b secondary
>>> sensor array b primary + sensor array a secondary
>>>
>>> (to have further redundancy if one of the instances stops or gets 
>>> maintenance)
>>>
>>> I wanted to keep the standard seasons skin, aswell as install the 
>>> neowx-material skin for each instance. BUT:
>>>
>>> After installation of neowx-material it already needs around 38 seconds 
>>> to generate the report files for one single instance. The ftp uploads 
>>> already are in 1 to 2 minutes too late for this instance, on both the 
>>> seasons and neowx-material skin. Tried it with the sdr-instance, the 
>>> interceptor-instance (with just the seasons skin until now) is less 
>>> affected by this.
>>>
>>> Is my neowx-material config somehow broken or is this a normal beaviour 
>>> for a 1 minute reporting interval with the neowx-material skin? 
>>>
>>> I even wanted to add belchlertown and weather34 in the future but I 
>>> guess that will be just too much, as it seems they will consume even more 
>>> resources, with the live reporting possiblities?
>>>
>>> Is the performance primarily cpu or memory (sd card) dependable?
>>>
>>

-- 
You received this message because you are subscribed to the Google Groups 
"weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/weewx-user/1b8d0e21-e7f6-4046-896d-821f2d971b93n%40googlegroups.com.

Reply via email to