lil52d ago
My #gotosocial alt has shit the database, somehow a database inconsistency can destroy your entire server.
Show thread0ct0pu52d ago
@lil5 nothing happens without a reason, I manage several platforms based on #gotosocial and it is unlikely that something will fall apart by itself. #gotosocial may not be perfect yet in terms of features, but it works flawlessly for a small community. Have a nice day ;)
Show threadlil52d ago

@0ct0pu5

I have never touched the database before this, the only way I interacted with gts is with a mastodon app or web client. Sure I have watchtower updates on, in the night. But nothing here says reason by administrator, rather it’s much more likely to be a bad fallback that’s not caught in code.

Show threadamd2d ago

@lil5

Did watchtower run an update? Did it get killed mid migration?

@0ct0pu5

Show threadlil5
@amd @0ct0pu5 Just checked the watchtower logs, it's not that: last gts update by watchtower was 2026-03-06
Apr 5 at 7:58amWeb
Show threadlil51d ago

@amd @0ct0pu5

Used an old backup from 2026-03-31 and the server won’t start, xD maybe it was watchtower after all. tbc

Show threadlil51d ago

@amd @0ct0pu5 Came back 12h later and that backup that wasn't working has magically fixed itself, so I can honestly say it wasn't watchtower 😝

I guess #activitypub just need some time to sync with an old backup.

Show threadamd1d ago

@lil5

Whatever it was. I’m glad it’s working.

@0ct0pu5

Show threadlil51d ago

@amd @0ct0pu5 Same the lesson I'm taking away from this is to use resticprofile everywhere I can, it's great!

I swear, home servers are the most dangerous a project can run on: shitty secondhand hardware, spotty internet, unattended updates, electricity cuts. And I love it.

Show threadamd1d ago

@lil5 backups are definitely key.

I run my GtS on cast off enterprise hardware with dodgy power and internet… a UPS and loads of backups are key.

Though I don’t do unattended backups. I do all those manually.