> I thought I would save myself the trouble and just restore from the > host provider backup from before the fault, but I got the same fsck > failure.
This makes it sound as if the provider is just doing a snapshot of the live file system, with all open files and unwritten cached data missing as it has to be when just cloning the underlying device, so this would explain why you need to run fsck when booting on the "backup", since it is very much alike when you just pull the power plug on a computer while its running. I don't think you had corruption per se, but rather both the outage and the "boot-from-snapshot" will give you similar/same experience on the next boot, all the unwritten stuff is missing and an fsck is required to fix up the dangling parts. I would also suggest doing other kinds of backups of the SQLite database, with a set of options described here would work: https://oldmoe.blog/2024/04/30/backup-strategies-for-sqlite-in-production/ so that you can rebuild even in the case where outages or the host snapshots for some reason will not restore service in an orderly fashion. -- May the most significant bit of your life be positive.
