Btrfs might just become “the ZFS of Linux” but development has faltered lately, with a scary data loss bug derailing RAID 5 and 6 last year and not much heard s

Btrfs might just become “the ZFS of Linux” but development has faltered lately, with a scary data loss bug derailing RAID 5 and 6 last year and not much heard since. It was not a per se data loss bug. It was Btrfs corrupting parity during scrub when encountering already (non-Btrfs) corrupted data. So a data strip is corrupt somehow, a scrub is started, Btrfs detects the corrupt data and fixes it through reconstruction with good parity, but then sometimes computes a new wrong parity strip and writes it to disk. It's a bad bug, but you're still definitely better off than you were with corrupt data. Also, this bug is fixed in kernel 4.12. https://lkml.org/lkml/2017/5/9/510 https://lkml.org/lkml/2017/5/9/510 Update, minor quibbles: lacking in Btrfs is support for flash Btrfs has such support and optimizations for flash, the gotcha though if you keep up with Btrfs development is there have been changes in FTL behavior and it's an open question whether or not these optimizations are effective for today's flash including NVMe. As for hybrid storage, that's the realm of bcache and dm-cache (managed by LVM) which should work with Btrfs as any other Linux file system. ReFS uses B+ trees (similar to Btrfs) XFS uses B+ trees, Btrfs uses B-trees.