Backup is growing rapidly without a special reason - from 2 to 4.1GB in 12 days

Configuration
,
1

It seems like the generated tar archives (full backup) grow a lot recently without me having done to my HA setup. From what was “once upon a time” 700MB on average the files grew to 2GB 12 days ago and 4.1GB yesterday.

Somehow I lack a tool to analyze the backup properly, so I can tell only the obvious differences of the files. So far I found out by opening the generated tar file that the contained file core_mariadb.tar.gz is causing most of the growth. I found found out that data/databases/homeassistant/states.ibd is now 12410MB big compared to 4700MB 12 days ago.

I had an idea that the recorded camera streams are stored in HA. But I set up my Reoling to record to SD card. Nothing altered in HA. So no clue again…
Can anybody help tell me how to find out what is causing the growth?

My setup: I use a supervised HA setup with HA and ESPHome and everything else updated to the latest version. The HA Google Drive Backup addon is working fine. Recently though I was told by the addon that my Google drive cannot hold the 5th backup.

2

MariaDB is your recorder and the typical cause is a sensor that reports lots of different values or maybe even unique values, like timestamps.

Check and control your recorder settings.

1 Like
3

See also: How to keep your recorder database size under control

4

I changed no recorder setting. The kind of major change was that I deleted three or so ESPHome devices and got different entity names. As far as I was told this name change does not matter as the entity id is used in the database .

Is there any way to access and analyse the MariaDb database within the backup?

5

I think the entity name is the entity id.
Friendly name is not the entity id though.

6

Also there is two purge functions to delete data from the database

7

You can use the phpMyAdmin addon (or standalone docker container) to poke around the MariaDB database. But you’ll still need to know SQL query language to do anything advanced.

I would expect your database to stop growing after you’ve reached the recorder purge setting (10 days by default). If your database stops growing and stays the same size after that, then you likely added or reconfigured a noisy sensor (or sensors) 10 days prior.

1 Like
8

Just a few words of analysis.

I am still not at a point where I see the problem solved as I still keep an eye on the backup growth.

Removing ESPHome devices from the integration and adding the devices again led to renaming of the entities. The database growth occurred as the recorder filters were not changed.

What I did so far:

  • solved the ESPHome naming problem by removing a device specific prefix for each device
  • changed the recorder filter according the new entity names
  • purged the database

So far the database grew 600MB to 900MB in 15 days.

I expected growth, but cannot tell

  • whether it is still within the limit
  • when the growth will stop due to regular cleaning of the database by HA (recorder settings based)

I shall keep an eye on the backup and adjust the recorder settings if necessary.

9

Thanks for the hint!

First I did not know how to access the DB with phpMyAdmin as I am no longer a day to day user, but then I dawned onto me where the access data is ‘hidden’.