Good morning, with the 2025.1.0 update, I noticed that the database grows significantly more compared to the previous version. With version 2024.12, I was at 50MB per day, but with the new update, I’m at about 120MB per day. I haven’t added any devices or sensors. Has anyone else noticed the same behavior?
Yes:
Even with the 2025.1.1 update, the behavior is the same. What can be done? Should a bug report be opened?
That’s neat. Do you have a write up on the SQL sensors you used for the breakdown?
Looks like your attributes are the main cause. Next step would be to dig deeper and see which ones.
Yes, I’d say so as it is not just you. Though it could be due to an integration you two share in common that I don’t use.
I only have the trend of the database size. The flat spot is justified by the purge data and the settings I have on the recorder, and it is correct. However, as you can see, since the January update, the slope is much steeper.
How can I investigate and find the integration that causes this behavior?
many thanks
There was a database-schema update (47 → 48) in the 2025.1 release. Maybe there is an low level convertion started that (hopefully temporary) is using more space …
I am going to divve deeper with mysql worbbench.
Will let you know my findings.
@tom_l do you know anything about that schema update?
I will post the sensors for the graph.
You guys are obviously more experienced than I am!
The only integration I’m concerned about is Local Tuya, which I use for some power meters. By any chance, do you use it too?
Yep, I have many tuya devices, and I use Local Tuya. But most of the tuya devices are using zigbee protocol. For zigbee i use zigbee2mqtt.
The rospogrigio one or xZetsubou?
I recently switched to the second one.
You might be on to something. I don’t use Tuya.
The PRs associated with this issue are the only db changes I’m aware of: Poor SQL performance finding start time state/stat results in HA Blocking Indefinitely · Issue #132865 · home-assistant/core · GitHub
I reverted back to rospogrigio’s original Local Tuya, and the situation seems to be back to normal. I have only one device missing, which forced me to switch to the xZetsubou fork. I’ll keep monitoring.
xZetsubou version
So I dived deeper, and this is what I saw:
After updating to 2025.1 there where devices that got more attribute-information. (also hidden attributes …??)
And that is now stopped…
I have to say that I also updated to the new zigbee2mqtt version 2.0.0.
And that was an breaking update.
So it took while to fix that one. And after that all data seems OK again.
So in the end, it looks like a temporary update-behavior on attributes, which causes also a states increase.
The database should now shrink after the retention time (in my case 10 days)
I keep you updated (when i do not forget that 
)
Have you always used the xZetsubou version of LocalTuya?
If you can, check that too, because I switched back to rospogrigio’s version, and now the database seems to grow as before (+10MB in 4 hours, compared to 60MB in 4 hours with xZetsubou).
I looked at some Local Tuya devices, and they seem OK.
Meanwhile ther graph is flatten again:
Now waiting for reduction …
As promised: the now shrinked database:
Almost back to the values previous to januari 2nd, just some little increae due to new sensors.
1 Like