b3141 - Sunshine hours & ET

[jon_iz]

I run both a RPi4 and PC via WLL and Weatherlink pro subscription. The RPi4 is a test platform, ftp'ing to it's own set of folders on my host.

I've noted from the release notes, there is fix for sunshine hours and ET should now work subject to Pro sunscription..
I have updated the RPi4 from b3133 to b3141 today and have noticed the following:
On the charts pages, daily rain and sunshine hours are not displaying data before the 27th July. If i roll back to b3133, the data is there on the graphs.
Is this intended?
https://duncreggan.net/cumulustest2/trends.htm#sunhours

Compared to the PC version:
https://duncreggan.net/cumulusmx/trends.htm#sunhours

The ET value on the first startup, reflects the value that weatherlink is showing for today, on the my index.htm (Now page)
Subsequent restarts of CMX, this value seems to reset to 0.2mm:
https://duncreggan.net/cumulustest2/index.htm

[mcrossley]

I run both a RPi4 and PC via WLL and Weatherlink pro subscription. The RPi4 is a test platform, ftp'ing to it's own set of folders on my host.

I've noted from the release notes, there is fix for sunshine hours and ET should now work subject to Pro sunscription..
I have updated the RPi4 from b3133 to b3141 today and have noticed the following:
On the charts pages, daily rain and sunshine hours are not displaying data before the 27th July. If i roll back to b3133, the data is there on the graphs.
Is this intended?
https://duncreggan.net/cumulustest2/trends.htm#sunhours

Compared to the PC version:
https://duncreggan.net/cumulusmx/trends.htm#sunhours

Nope not intended. If you look at your Historic charts you have several large gaps in your sunshine hours data, one of which ends on 26th July. I'd have a look at your dayfile for any issues. If you cannot spot anything then perhaps post it so we can take a look?

The ET value on the first startup, reflects the value that weatherlink is showing for today, on the my index.htm (Now page)
Subsequent restarts of CMX, this value seems to reset to 0.2mm:
https://duncreggan.net/cumulustest2/index.htm

I'll take a look...

[jon_iz]

I've looked at the dayfiles. You are correct, the sunshine hours have all reset to 0.0 prior to 27th.
Exactly the same thing happened on both the RPi4 and the PC - but i rolled the PC back to 3133 - both day files are attached.
(I forgot to image the RPI4 first, but that's not so important as it's a test platform anyway, so i probably need to try and roll the data back)

My upgrade process was to over-write the existing install with the new release and run it up, so it looks like it's a glitch in the update.
The log file for that initial run is also included.
Both platforms obviously run independent of each other, with the common factors being the data sources of WLL/Airlink and Weatherlink pro subscription.

ET:
CMX on the RPI has not been restarted today, and CMX and Wetherlink are broadly tracking ET figures.

WL ET.PNG

CMX ET.PNG

So i wonder if the ET figure was read just the once and then it's reset with each start rather than appending to the cumulative figure?

[mcrossley]

I'll take a look at the ET later.

Regarding the dayfile. Have you used the dayfile editor? That is the only time Cumulus will change entries in the dayfile that I can think of.

[jon_iz]

I'll take a look at the ET later.

Regarding the dayfile. Have you used the dayfile editor? That is the only time Cumulus will change entries in the dayfile that I can think of.

There's a yes and no answer to that.
Yes on the PC instance data around 19th July, when i discovered false rain counts (The ISS rain funnel had a blockage and was holding water and dripping on days when it clearly wasn't raining).

I did not update the data on the RPi instance - so this is not the cause. (The RPi instance was created sometime in June, based on the data and amended config from the PC instance)

(you can see this where the rain blips are removed mid june here https://duncreggan.net/cumulusmx/trends.htm#dailyrain and exist here https://duncreggan.net/cumulustest2/tre ... #dailyrain

[mcrossley]

Sorry, I have no idea what has happened to your sun hours. The only time Cumulus alters anything in the day file is if you use the editor, and then it rewrites the whole file for each edit so if there was a problem with the sun hours then might expect it affect all the records.

The ET issue will be fixed in the next release.

[jon_iz]

Not to worry, a restore of backup data to a point earlier in the month and a resync with weatherlink servers seems to have resolved the issue.
I'll try again with the PC platform and repeat the restore option if the same thing happens again.

[jon_iz]

Just to close off the sun hours - upgrade to b3143 didn't exhibit the same issue on either instance.

ET still resets to 0 on the web index.htm with a restart of CMX, rather than pulling back the cumulative figure for the day.
Perhaps this is a tag issue on the webpages on the website, as querying the API with x.x.x.x:8998/api/tags/process.json?ET returns the correct value after a restart?

[mcrossley]

That's odd because the web site and the API both use the same (only) variable that the current ET value is stored in.

I would expect zeroing on the first start of the new version of MX, but thereafter for it to be OK.

[jon_iz]

That's odd because the web site and the API both use the same (only) variable that the current ET value is stored in.

I would expect zeroing on the first start of the new version of MX, but thereafter for it to be OK.

You are of course correct.
After update, leaving running for a day, switching off for few hours and catching up does show the correct value.
So I will take a