Btrfs in 2026: Ready for Production (Mostly)
Btrfs has matured dramatically. Facebook uses it on millions of machines. openSUSE and Fedora default to it. The filesystem is stable — but some features still carry warnings. Understanding which is which is the difference between loving Btrfs and losing data.
What Works Brilliantly
Snapshots are Btrfs's killer feature. They're instant, copy-on-write, and consume almost no space. btrfs subvolume snapshot /home /snapshots/home-$(date +%Y%m%d) creates a point-in-time copy in milliseconds. Combined with snapper (automatic timeline snapshots), you get the equivalent of macOS Time Machine on Linux.
Subvolumes let you partition your filesystem without actually partitioning. Separate subvolumes for /, /home, /var/log, and /var/lib/docker means you can snapshot or rollback each independently.
Compression is transparent and fast. zstd:3 reduces storage by 25-40% with negligible CPU overhead.
What's Still Dangerous
RAID 5/6: The write hole still exists. If power fails during a RAID 5 write, you can lose data. Use RAID 1 (mirroring) or RAID 10 instead. The Btrfs wiki still warns about RAID 5 in production.
Swap files: Only recently stabilized (kernel 5.0+). If you're on an older kernel, use a swap partition.