I setup my HA about 18 months ago, and when I viewed ‘History’ I could see historic data all the way back to the beginning.
Checked today (2nd Jan) and now see data prior to 23rd Dec 2025 as ‘Long Term Storage’.
I haven’t intentionally changed anything related to storage, so what has happened ?.
( System is running on an RPi 4, with 44Gb of free space. )
What is a value of purge interval of recorder in your setup?
A default value is 10 days.
History view shows all data before that interval as long term stats (if stats is generated for this entity).
Where do I set this ‘purge interval’ ?. I’ve never tweaked anything to do with historic data before, it just worked as I wanted it.
If you did not change the purge interval then you only ever had 10 days of state data. All the older values are/were long term statistics. This is the way Home Assistant has worked for a lot longer than 18 months.
I can assure you I used to get a lot more than ten days on my temperature chart, but now it won’t show prior to 23rd Dec. Same with the pump monitor chart.
I chose HA because it was not cloud-based, and it fitted well with my Switchbot temperature devices. However, I have found that it needs a lot more geeky intervention than I’m capable of, and I’m losing faith in it !
You just said that you still can see data before 23.12 as “long term stats”. This is what you probably saw before.
Long term stats here are statistical data calculated every hour. Suggest to read docs about what a statistics in HA is.
Short Term Storage keeps data in 5 min interval for 10 days.
Long Term Storage keeps data in 1h interval forever.
The same info (and much more) you will also find in the docs, if you know how to google.
If you don’t know, HA is probably not the best system for you. There are many less powerfull systems that might suits better to your needs.
There could be an issue with the sensor’s statistics. Go to Developer Tools → Statistics.
Find your temperature entity and see if there is a FIX ISSUE button to the right of it. If there is press it and read the issue description.
Also check your Settings → System → Logs for errors related to the sensor or recorder database.
‘Long Term Stats’ seem to be averaged or aggregated into 1 hour periods. My pump monitor generates between 800 and 1000 pulses over a two minute period. This occurs between 40 minutes and a few hours, depending on rainfall.
I could see the 2-minute ‘spikes’ going back for months. Now they are just blurred averages.
I don’t remember, but when I installed HA, was there the option to select a retention period ?. If this was the case, then it could maybe have reverted to default due to an update or some kind of glitch.
Whatever it was, I definitely had full historical data going all the way back to day one.
Nope. Home assistant never worked like that. It only kept 10 days of real time state data. The 10 days or older data was always averaged into hourly LTS. Before the introduction of LTS this data was simply discarded.
No this is not an option during onboarding. You have to deliberately alter the recorder integration to get a longer state data retention.
As we keep saying, this is not possible without deliberately altering the recorder integration.
And if you do this then know that keeping state data for longer than 30 days or so can cause performance issues.
I know you all think I’m mistaken, but I can see the difference between the display of the two types of storage, and I know what I was looking at was historic.
This is not over yet.
Yes that was introduced a while ago. Before that all you would have seen was the 10 days state data.
