Cumulus Support

I have several extra sensors, all on the same Ecowitt gateway as the main station. One of those is soil moisture, currently a value of 82% (easy to spot in the log). When I disable the sensor I expect the sensor to have a null value. Instead it keeps logging its last value (I can accept that) but it also logs a value some 24 positions ( ) further. The last log section shows (with two 82 values the sensor is disconnected/disabled): What is happening here and what should the moisture value be? null or 82? I think null would be best: it makes it compulsory for the user to do something and it shows the sensor is out of order (either not connected, out of power, disabled or whatever).

And I can add that restarting without the sensor active works as expected (a null value) when activating the sensor I get only the 82 value on that position, not 25 positions further. The value is not displayed in the dashboard however. I'll add a logfile tomorrow after restart of CMX.

A bug in the code, I'll fix it.

mcrossley wrote: ↑Wed 08 Jan 2025 10:36 am A bug in the code, I'll fix it.

OK. Thanks.
No need for a logfile I assume

No - but looking at what is required, this will take a big chunk of dev time

mcrossley wrote: ↑Wed 08 Jan 2025 11:06 am No - but looking at what is required, this will take a big chunk of dev time

OK. With improvement at stake I am a patient man

Digging into this some more, it turns out to be even more complex than a first look would indicate. It will add a considerable amount of extra processing to every data message received from the station

Unless AI can come up with a brilliant solution I haven't thought of! Framing the question to get any sensible output is the hard part!

Interesting that this is so difficult. I did not expect that.
And aha.... AI is entering the CMX environment How and which tool? Do you put the software as input?

mcrossley wrote: ↑Thu 09 Jan 2025 11:33 am Digging into this some more, it turns out to be even more complex than a first look would indicate. It will add a considerable amount of extra processing to every data message received from the station

Unless AI can come up with a brilliant solution I haven't thought of! Framing the question to get any sensible output is the hard part!

Extra processing for every message received... I thought that over.

Wouldn't it be more efficient to just check every 5 or 10 minutes (async) and array of available sensors for their presence (in the latetst communication with the station)) and those who failed get ignored as if not present after that. If I get a value for a failed sensor once or twice I could not care less. It is just that it has to be detected somewhere that a sensor is lost (and possibly raise a general alarm on a sensor lost indicating which sensor?). No need to do that every 4 seconds with every message of the station I think.

It's not AI but maybe it helps

The problem with that is you then have to record the time the last data was received for each sensor.

The big complication is sensor data coming from extra sensors on a secondary station.

Example:

My primary station has Extra Temp #1, my secondary station has Extra Temp #2.

The primary station sends its data, there is data for extra temp #1, but nothing for extra temp #2. But we cannot set extra temp #2 to null, because it may be coming from the secondary station. The same is true in reverse for the secondary station.

So the boths stations must record if it previously had data for each sensor, and if it "disappears" or becomes invalid, only then set it to null.

Further complications, Davis for instance send a fixed data structure, with an entry for each sensor, each sensor contains data or not. Ecowitt only send the sensors that actually have data. Simple in the past, Ecowitt MX just parsed each sensor and recorded its value. Now it must check if every possible sensor is present, and if not, was it there in the past? This all equals lots more code for all the station types.

I see. Thnx for explaining.