Just crash with faulty SD card. Had a snapshot from just 6 weeks back, so I thought no worries.
Tried to restore, restore and wipe and finally restore through SSH. No such luck. Non of my configurations are restored, only a few addons.
Manage to find my config files by opening the .tar packed file. Had to copy paste text from them into my new setup. Most of my system is Z-Wave, so that was still in tact on the stick.
Get me thinking that this restore option is at best risky, and I will not trust it from now on. Either I will image the SD card, or I will just backup the config files.
Anyone having other options for backup?
As the system is getting pretty large, I find it quite inconvenient when it goes down.
Lots of replies to this one I see. I don’t know what exactly constitutes a full snapshot, because until today, it meant I got a full restore. I guess that isn’t always the case. My string of bad luck today began with an eMMC go belly up on my odroid n2. The entire reason I found the odroid platform attractive to begin with was not having to deal always failing/corrupting sdcards. Not wanting to wait for eMMC modules to come back in stock again, much less the 2+ week wait for shipping from Korea, I had no other choice but to fall back on my RPi4. I happen to have an unopened SanDisk “high endurance” sdcard on hand. Balenaed the latest RPI4 64-bit image and fired it up. Enabled samba, copied over a snapshot that I downloaded using the web UI (not sure if there’s a difference here since it had a human readable filename). Anyway it restored my addons and little else. I dove into the archive file probably 3/4 of my yaml files had not been backed up at all in the snapshot. My previous snapshot is nearly a month old. A month is like dog years. I’ve been working like mad to learn this entire system and the changes have been large and many in the last month. At this point I should probably start from scratch and just copy/paste whatever yaml code I somehow managed to get right.
So are end users expected to conduct a thorough examination of the snapshot archive contents each time a snapshot is made? Because it would certainly seem like a wise move.
Did you possibly create your backup by calling the service “hassio.snapshot_xxxx” and not through the gui?
There is a very old bug on github on this subject that does not get any love.
Long story short: Manual backups almost always fail on the file “home-assistant_v2.db-shm”, so no files after this (in alphabetical order) will be backed up.
There is no way to know that your backups are incomplete, unless you either look in the backup-file or in the system-log.
In my opinion unreliable backups are very disconcerting but his opinion does not seem to be broadly shared.
No I always use the same procedure - purge/repack, then full snapshot from the web ui. Always name the snapshot YYMMDD_HHMM00 (fudge the seconds by using 00).
Putting together a daily automation to do all of this had been on to-do list. Well it’s 8 weeks of work that should go a bit quicker the second time arkum