The Ongoing BcacheFS Filesystem Stability Controversy
https://hackaday.com/2025/06/10/the-ongoing-bcachefs-filesystem-stability-controversy/
@hackaday I'm starting to wonder if modern next-gen filesystems are approaching an inherent limit of human ability to mentally model and manage complexity.
Microsoft failed so hard with ReFS that I literally had to go look up the name again. Btrfs is fine as a relatively simple, single disk filesystem, but becomes an untrustworthy mess if you push it too hard. Then here's bcachefs.
Why do we keep failing so hard at reimplementing things that Sun and NetApp already did?
@jimsalter @hackaday I worked on a Linux container manager. We kept hitting semantic rough edges in various COW filesystems to the extent that we revisited at least one filesystem over the course of a single year (it was an XP team, so the cycle was generated by bugs encountered rather than any more thoughtful considerations).
Ultimately, the file system was made pluggable, so users could pick their own poison.
So it's no surprise that a relative newcomer is unstable.
@jimsalter @hackaday ZFS is alien tech. The temporary open sourcing was a random winning lottery ticket.
@coder @hackaday OpenZFS isn't perfect either, and if we're not careful, low quality development could eventually corrupt its reliability as well. It took us YEARS to finally close that encrypted replication bug.
Software is never perfect, and I'm not trying to roast OpenZFS or its current dev community, but I think we have to stay realistic and cautious. It's still software, still under active development, and still vulnerable to accumulation of tech debt and error.
@jimsalter @hackaday sure! I never saw Stargate, but likely there are suitable sci-fi cautionary tales about alien tech we don't understand :D
But really, it stands out on being a few years ahead of its time, like some other Sun stuff. I'm not sure we will see many similar things happen.