Accounting for accidental deletion or sabotages
Hey all, just wanted to check in regarding accidental deletion or sabotage. If you're keeping the files in a bucket, you could of course "keep all versions", but the database entry is removed for these images.
How would you functionally recover your photo's in this case? Excluding backups (say it's aged out and you don't realise for years). Is there an offline decrypt script so you can take the photo object + encryption keys an end up with a photo output?
11 Replies
Remember to use /solved to mark your thread as solved once your question is answered.
Hey currently you do need the DB data in order to decrypt, but yeah in case of DR having such a mechanism would be good
Also I think there’s no limit to what if. Having backup and automated process to verify full/ random check is what will work for majority of the audience. Of course there can be more advance solutions related to this, but for most people just regular backup should be more than enough
Yeah I guess my issue is, if I keep all the data in buckets, then I'll have to sync them somewhere local to be able to back those buckets up (along with DB + keys of course).
Alternatively I could do full export of the raw data every night, but as the data grows, that becomes less possible. Especially if I'm hosting say also for family, then bucket backup itself becomes the only real option.
Trying to decide if migrating from immich -> ente makes sense, and the backup/recovery element is my largest hurdle
Valid concern but having an unencrypted backup which can be automated via cron job can be done, the only constraint will be storage in that case which is avoided in case of Immich since data's unencrypted
Exactly yeah. Just trying to weigh up the pro's and con's of each. Currently photosync from phone -> encrypted ZFS which is indexed by immich and backed up by duplicacy encrypted to B2
Hey @TickleMeWalrus can this be marked as solved?
Yeah I guess is it probably more appropriate to make a feature request via GitHub then? If that’s the case I will mark it solved
-# If your issue is resolved, you can use the /solved command to close the thread. If not, use /unsolve to cancel.
Thread marked as solved. It will be closed in <t:1750589521:R>.
And yes if it's a new feature or enhancement make a discussion on GitHub
This thread is now closed.