Two terminal disk tools that go together:
- ful — a live dashboard of your mounted filesystems: usage bars, used/free/total, and per-device I/O. Which disk is in trouble? Not an acronym — a suffix, as in watchful (the title bar shifts to plentiful, watchful, woeful, or dreadful with your worst disk).
- dum — short for disk usage monitor: an ncdu-style tree explorer that watches filesystem events and illuminates what's changing: directories receiving writes glow green with a live rate and sparkline, shrinking ones glow red, sizes update in place. What exactly is moving?
Install both with one command:
cargo install --git https://github.com/kek/ful
ful # default 2s refresh
ful --interval 1 # custom refresh interval (seconds)
Keys: q/Esc quit · ? help · a toggle pseudo filesystems.
dum # explore the current directory, watching live
dum ~/src # explore a specific path
dum --no-watch PATH # plain explorer, no live updates
Keys: arrows/hjkl move · ⏎ enter · u/⌫ up ·
s sort (size/rate) · r rescan · ? help · q/Esc quit.
dum is read-only: it never modifies, moves, or deletes anything.
- Sizes are allocated (on-disk) bytes, like
du. Hard-linked files count once per scan (by(device, inode)), so multiply-linked inodes don't inflate directory totals — again matchingdu. - macOS reports I/O per physical device; if
sysinfosurfaces no per-disk counters, ful's READ/s and WRITE/s columns show—. - dum's live layer goes through
notify, so FSEvents on macOS and inotify on Linux. CI runs the whole suite on both, watcher test included. Windows is neither built nor tested. - If event watching fails or overflows, dum keeps working as a plain
explorer and shows "degraded" — press
rto rescan.