Is this sensor a total consumption-sensor or one that resets every day? If it is the latter, than that is probably the reason.
Is there a trick - it says I can only add one screenshot - should I post 6 replies now or create a PDF ?
Try posting the images now.
Alright - this is my current dashboard:
Grid Import Sensor History looks good:
Grid Export:
Solar Production:
Watermeter:
All sensors are MQTT sensors configured as Total Increasing
Only Grid Import and Export are coming from the powermeter over modbus but also configured as Total Increasing.
modbus:
- name: hub1
type: serial
baudrate: 9600
bytesize: 8
method: rtu
parity: E
port: /dev/ttyUSB0
stopbits: 1
sensors:- name: sdm630_total_power_import
slave: 11
address: 72
input_type: input
count: 2
precision: 2
data_type: float32
unit_of_measurement: kWh
device_class: energy
state_class: total_increasing
scan_interval: 1800 - name: sdm630_total_power_export
slave: 11
address: 74
input_type: input
count: 2
precision: 2
data_type: float32
unit_of_measurement: kWh
device_class: energy
state_class: total_increasing
scan_interval: 1800
- name: sdm630_total_power_import
So I double checked again - the history graphs of all sensors are correct - only in the dashbboard it shows wrong - I think there the statistic values are used - they are wrong - why I do not know:
So here for examle - Grid Import - all of the sudden at 4:10 am - the import was minus 64,34 - so a negative value…
Also the solar production (the PV is offline during night) the statistic shows at the same time a negative production:
Can you pls help me where I find it ? I have not added this sensor manually as it was auto discovered:
Never mind. What does its history look like?
Last night only:
The History Graphs all look perfectly fine- only increasing- no negative values or peaks…
Any idea what that coule be ?
It happens almost every day but not all days…
Is there a way to delete maybe ALL statistics if that is causing the issue but not the histories ? The histories all are fine…
What I also found out in the DB at that time - the sum of changes to 0 and seems to restart to count from there…
Then I can’t explain it. Statistics are generated from this state.
It is very very strange but I am not alone → Statistics reset to zero every now and then - #9 by PaStaDA
@sedael has the very same issue obviously 
I saw today somewhere that at 04:10 the database is purged automatically, maybe that has some relation?
LTS are not purged. Only state values and events.
Still noone with an idea why that happens ? It really happens almost every night now and with all “Energy” sensor - so now even my watermeter reports every night that I have “exported thousands of liters to the grid” Can I somehow delete the whole LTS and also if needed the history - I am now so frustrated that I even would delete all data from the last years to get a clear picture - with these wrong values in the dashboard, it becomes totally useless
I have the same problem. I have nothing changed the last few months except updating home assistant and suddenly it behaves like this so i am pretty sure that’s another bug.
@PaStaDA and @Zahnpasta1
Just a thought from a novice. Maybe the database purge isn’t working correctly. 
You could try TEMPORARILY disabling the purge to see what happens. See:
Recorder Integration
FYI I use MariaDB rather than the default SQLAlchemy. Also, lot of folks using RPi’s have issues with SD card memory. Here is an article on Maria DB:
Optimize Your Home Assistant Database
hey there,
I’m also getting wrong values almost once a day. Maybe 2 days of 7 are fine.
The wrong values in my diagram are the total values from my inverter not the diff. So maybe there is sth wrong in specific circumstances and the diff won’t get calculated?
I can’t find any logic in the timestamp the wrong value is recorder…
Sometimes its twice a day.
as you can see, I’ve got a qcless inverter. I’m runnning the solax integration…
thanks for any help
Hello,
I have similar problem with energy dashboard. Every day several minutes before or afrer midnight I have invalid value at only one device. All other devices are correct. All values are read from same device via modbus. Entity history is normal, no incorrect value. I guess problem rised after HA update to version 2024.05. There was no another change in HA or connected device. Thank you for any advice.
Tomyc
This may help