Recently I have been having a few internet stability issues late night / early morning and had no reason to think this was related to home assistant.
However after making a mess of my yaml in an attempt to split the configuration and update all my templates to the new template format, for the first time ever I resorted to restoring from a backup.
I take nightly partial backups and weekly full backups that are offloaded to Dropbox. When looking through my backups I noticed that they had jumped from approx 350 mb to nearly 4 Gb, not gradually increasing by literally jumped.
Now I can explain why my internet was becoming unstable at night as clearly I was uploading nearly 4 gb every night.
It seems that my partial backup automation (that’s been in place for years) that had never had the media folder included, now all by itself had the media folder included. This seems to have coincided with the recent change to media. But goes unexplained as to how this has happened as I have never changed or even edited this automation since I set it up.
However after restoring from the previous days backup I was a little concerned by multiple errors and logs relating to recorded and database not starting etc. No amount of restarting home assistant and or host cured this issue.
It seems that a previous known problem existed relating to corruption of the DB if backing up while the DB is running. However some posts suggest this issue was fixed.
My experience would suggest it’s not. Apparently the potential way round this is to stop the DB before making a backup, but this is messy as not only does it cause massive list of errors in the logs but there also appears to be know why of determining when the actual backup has completed and therefore would be left having to set an unknown time frame before starting it back up.
So as home assistant is clearly making massive attempts towards making itself more user friendly and available for the masses, surely this whole backup thing needs to more love. I was initially amazed that this could happen and equally if not more amazed that there is no way triggering of a backup completion in what is essentially a state machine.
I absolutely love home assistant and support it on every way I can, but this is the first time ever I have felt truly let down by it.
So would urge people to check their backups just in case and would love some ideas of fail proof ways of backing up the DB.
Rossk