Apparently there is no regression test done on Home Assistant, it seems to be a toss up if the structural changes implemented really works.
That’s not that big a problem, as I do full HA backups every night, or at least I thought so, because the ‘partial restore’ that I wanted to use for this, doesn’t restore the config, so what good is it?
Appdaemon, config gone, fortunately I do a lot of backups, and using proxmox, so I had a working config I could restore from that.
Appdeamon is really essential in my setup, as it handles all buttons.
No this bug (which also absolutely HAMMERED Node Reds addon) condition was missed in the testing per Frenk himself.
It was tested and missed. It happens. This one just had some significant consequences. Glad you had your backup. Keep doing it and keep testing your backups regularly.
Ive had that happen to me in my career too. (mine ompletely bricked 2000 desktop workstations and had to spend 72 hours straight rebuilding them)
If you read the dev notes in the various PR entries it is a good thing they’re doing. It improves security and eases maintenance unfortunately it was one missed thing that caused it to blow up.
Im less worried about that part because accidents happen. It wasn’t malicious.
Im concerned there’s some consequences that were missed from the change. Such as:
If I were using NR or AD (the two most prominent addons affected) and all of a sudden I couldn’t have read write access to the full HA namespace… And say I was dumping camera snapshots or JSON files down to the disk to be picked up by a core process…
Where do you drop the files? There’s no common folder you can write to anymore except a few small places. (at least according to the disk map in the current docker container…)
Things like that seem to have been missed im already seeing two issues directly related to the move that wasn’t ‘my config got wiped out’ we know how to fix that now.
Transfer folder? Hmmm.
My NodeRed was oddly not gone, it survived (thank god that would be another, much bigger, disaster).
It’s not a blame game, so I’m not blaming Frenck in any way, for me this is more of a post mortem analysis and a real question of if tests are done.
The issue is compounded with the inability to do a partial restore.
The problem as I see it is that there are not a procedure set up for these kind of changes, which might have caught it, with the growing complexity of Home Assistant, small changes can have very big consequences.
1 Like
Ironically enough this change is intended to help fix it.
It ultimately brings the ENTIRE config of an addon into one place so we don’t need to do the gargantuan efforts to resolve it as we all did.
In newland your addon backup would have been enough. Its the pre update config that caused the restoration issue you mentioned 