"Recorder: Purge Entities" not working at all

Configuration
1

Hi !!
I’ve had some issues with a huge database even using the recorder integration with include option, selecting the entities of interest (mostly temperatures, humidity, automation histories, …).
I had to delete it.
Moved to MariaDB without any issue.
Added some more features.

It looks like this :

recorder:
  db_url: !secret db_mysql
  commit_interval: 30
  auto_purge: true
  purge_keep_days: 7
  include:
    entities: Zigbee stuff

The database is growing at about 80 MB/day.
I have noticed one Zigbee device generating a huge amount of data ( TS0601_smart_air_house_keeper with 7 sensors. At a point that I decided to disable the recording for 6 of them till I find a way.

I tried to call a service : Recorder: Purge Entities. But it does nothing. No trace in the logs too.
The database continue to grow.
Does this function work ?

Thanks :slight_smile:

2

That’s not much. Mine grows at 350MB/day, until the purge keep days is reached (7 days) the it levels off at just under 2.5GB. I use extensive excludes.

So yes it does work.

Try setting the repack option to true when doing a manual purge.

3

mine reached 6.0 GB with extensive include :slight_smile:
Is there a way to purge an entity ?
I can’t manually remove anything from phpmyadmin on my SQL database

4

For how many days?

5

There was no limitation. If I’m not wrong, the max is 10 days right ?

6

Yes 10 days by default if you have not set it. Do you need 10 days?

7

I have set now 7 days… It is to monitor the temperatures and my heater to optimize the settings.
I have started to move entities to InfluxDB as it may be less space consumming

8

That should drop it to about 4.2GB after 7 days.

9

I have removed 6 sensors that were recording about 20,000 entries in the DB. It should be better too.
Do you know if there is a possibility to reduce the interval between 2 records ?

10

Well, I can’t explain why I see it now, but the purge is working now. I see it on myphpadmin.
Just did the last update of HA… could be a coincidence. Yesterday it was not working