Egregoros

Signal feed

Timeline

Post

Remote status

Context

13
BTRFS is the most german file system. So many weird autistic issues, always trying to be better and never delivering, sounds good on paper but sticks a knife in your back when you aren't looking.

Just like real germany. (This has nothing to do with the fact that SLES uses it as the default file system.)
@meso @phnt to be fair i haven’t seen much data loss with btrfs. in that regard it has been far more reliable than ext4 in my personal experience. the trouble i ran into was performance degradation that rebalancing didn’t fix, and the fs just completely seizing up because it ran out of metadata space. couldn’t free it up by deleting files or snapshots, none of the tools and recovery instructions worked. in the end i added an in-memory loop device to the fs to fix it.

i still think btrfs is basically an elaborate prank by oracle engineers. both its development and its adoption went far too quickly, and it didn’t take long for it to get bigger than even XFS. all that code was merged and deployed in production without much testing or design iteration and while both documentation and tools were severely lacking.

but that’s not german, as much as i hate the german flavor of tech autism. that’s how silicon valley operates.
@phnt @meso tbqh i don’t know what demand btrfs is even trying to meet these days. it’s barely good enough for hobbyist data hoarders and not really designed to run on finite storage that is used to near-full capacity, and Big Enterprise strongly favors object storage over traditional file systems for obvious reasons

Replies

2
@lanodan @meso @mia Also stupidly easy to integrate with package managers, so you can easily recover from bad updates.

One annoying issue at least with snapper is that the rollback functionality is hard coded to the opensuse layout, so it is useless outside of that. But it's not hard to find the snapshot you want and rolling back is done with `btrfs subv create` since snapshot is just another subvolume.