A graph of my disk percentage use is below. Suddenly at midnight this morning, went a LONG way down. Now I am quite pleased because it was starting to worry me. So why did it go up steadily, and then suddenly down? Supervised on debian 11.
Well, look at the tar before and after to know
You’ve got me curious, too. If a change made something more efficient, it would be good to know what that change was. If you get a chance to poke around in the tar, please let us know what you find!
Thank you for encouraging me to go further. I had never opened a tar file before because I’m on a windows machine. I took a crash course in it and I’ve identified the file that caused the drop:
I don’t know what this file is. Is this the actual home assistant guts? if so, then something they did in the update made things more efficient. If not, I’d need to crack it open to see what’s different. Should I crack it open or does somebody know what the file is?
It’s just a directory (“folder” in Windows). You need to open it up and drill down until you find the big difference in size.
From what I can tell, this was an HA improvement not a change from me.
The file structure changes significantly between the two backups and the packed size changes. But when you drill into all the individual folders/files, the difference in size is minimal:
Upon closer inspection, there’s a DB in one and not the other. I had been noticing it in my files and it looked like it was some sort of corrupted DB but I was too afraid to touch it. Looks like maybe they ran something to purge it?
This also stands to reason with the OP. If there was a corrupted DB underneath, then there was another non-corrupted DB still taking in new info and growing each day. Then, when corrupted DB purged, size of file immediately dropped.
Also, I do recall accidentally restarting HA while in the middle of a backup sometime in the last month. Wondering if that’s how I got a corrupted DB.
I think I saw a commit that explicitely exclude those from the backup indeed.
They are still on your filesystem, though, and can be removed if your HA is currently running fine with a more recent DB.
I took your advice because based on the date (january) I was confident that it could be discarded. I removed the file and HA really didn’t like that. But what I determined was removing the file bottomed out performance and it killed a bunch of processes for several minutes. HA eventually came back up and I was able to determine what happened by looking at my host (it ran out of memory). I have confirmed removal of the file and everything is working fine now. Thanks for all the tips everyone! I was just randomly curious because I saw the same thing the OP described and it turned into a healthy rabbit hole learning experience.
None of this helps Nick ans he was not talking about the size of his backup files.
Then again I don’t really care beyond satisfying curiosity. The space was used up, then it wasn’t. Mystery unsolved, but not keeping me awake at night