Due to the number of sensors and sampling frequency, in my setup many writes are sent to the home assistant database by the recorder component.
At the same time, all sensors are logged also to Influxdb for graphing via Grafana: IOPS and wear are more or less doubled.
Together, the load on the disk (both IOPS and overall wear) is higher than I would like.
The following question is general and not specific for my hardware/system, of course I will evaluate which values should be excluded, this is not the topic
My question which I could not find answered elsewhere is: what are all the implications of disabling the recorder component altogether?
“Disabling” can either mean excluding everything, or removing the default configuration and adding back every component except recorder, logbook, history.
“Disabling” doesn’t simply mean deleting the recorder: entry from the config.
Related question: what are the implications of setting a short retention time and frequent purges for the database, when the recorder is not disabled? Which features am I going to lose for data prior to the last purge threshold?
History graphs won’t work anymore, and logbook neither, this is obvious. What else?
What kind of native automations won’t work?
And if I use Node Red to develop some automations, will these be affected too? I haven’t touched it yet so I have no idea whether it queries HASS or influxdb directly.
Among these, which lost functionalities can be compensated by querying Influxdb directly (via Influxdb.sensor) from within HASS?
If you remove recorder: from your config, you get a default implementation. That’s not likely what you want.
I did briefly explore using the same Influx database for the integration and the sensor. That turned out to be a mistake, I’m guessing that it sent an updated value to InfluxDB, the sensor read it from there, at which point it got sent to InfluxDB again…
The right answer is to exclude entities that are noisy from recorder.
You could also move the database to a USB attached drive, and do the same for InfluxDB.
For the purpose of avoiding burning out an SD card, the purge interval doesn’t help much. At the end of the day you write it, then erase it. It doesn’t matter if that happens with an interval of hours, days, or even weeks.
The core of HA will still work, as it’ll store the state of things internally. Sensors and switches won’t retain their state across restarts though, and obviously some UI features may not work as expected.
But can I for example write automations that get triggered if a sensor is within a certain range for the last X minutes? Is this something that requires knowledge of the prior values and therefore impossible once history is gone.
As long as home assistant is up and running, that information is stored in the states machine. It has nothing to do with recorder. You can test this pretty easily.
