The sensor for 1 day not resetting properly

[protik77]

First, thanks for the integration! It's been working nicely so far for me.

Every 2 days of so, my sensor for 1 day is not resetting at 12 am. As a result, I am seeing a spike at around 12 am. I have attached the screenshot from the energy dashboard and sensor data around the time. Can you please look at it when you have time?

Energy dashboard:

The sensor providing data:

[pekkeri]

I am having similar issues frequently.

For some reason the sensors sometimes keep giving data several minutes past midnight (more than the 1m45s spec in code?)

[jeffutter]

I'm experiencing this too, glad to see I'm not alone. I might try to create a integration rollup based on the 1m sensor as a stop-gap.

[jpearce73]

I just took screenshots that look identical to this and came here to post the same issue! The interesting thing is times of the glitches look very similar. Resets at 00:01 Glitches back to the old value at 12:03 Resets again at 00:19 And I confirmed all the sensors do this, the mains and all circuit sensors I will try and turn on some debugging information with the logger and see what it says, obviously, we need to determine if the data returned from Emporia is at fault or if its the integration.

[DarkStar1973]

Same here

[wirecuttr]

This is year-old issue, but just posting that I'm also seeing the same issue on occasion. I noticed it because it does mess up the Energy dashboard as noted in the original post. Are people just creating their own utility meters to work around this?

[magico13]

The drop to zero is code I've got on my side to force it to reset at midnight, the spike back up is from the API returning old data. That API is queried every 15 minutes, so the width of that spike should be about 15 minutes wide unless the API returns old data for multiple requests (which can and does happen). As long as it drops back down by 1 AM it shouldn't cause any long-term issues with the energy dashboard as the statistics should see it as a large increase followed by a similar decrease in the same hour. That is what happens for me at least, as you can see by my data for March 11th.

The fix on my side would be to ignore the data from the API when a big spike is detected, while not accidentally throwing out good data from heavy usage or errors in how I integrate the minute data. I mostly only have the immediate data to look at, not the historical state, so looking for a spike isn't trivial. And what constitutes a spike may be different from one person to another, for instance someone might use 10kWh in a day and I can use more than that in a half-hour at times so a simple "jumped up by more than X kWh" might not be sufficient.

You can actually watch this happen in the Emporia app, if you see a big spike happening just after midnight and you open the app, the Day data will still show the numbers from yesterday+today instead of just today. If you look later at the graph in the Emporia app it won't show that error because they rebuild the graph each time from the raw data, whereas Home Assistant builds the graph from the data it got from the API at that time, so you have to look while the API is still giving bad data.