Menu

Find the largest files on disk

Updated 2026-07-05

Problem: Disk space is running low and you need to find out what's using it.

Solution:

du -ah /var | sort -rh | head -20
2.5M	/var
2.0M	/var/cache/big.dat
2.0M	/var/cache
504K	/var/log
500K	/var/log/syslog
8.0K	/var/small.txt

How it works:

  • du -ah /var prints the disk usage of every file and directory under /var, in human-readable sizes (-h), including individual files, not just directory totals (-a).
  • sort -rh sorts that output by size, largest first (-r reverse, -h understands human-readable sizes like "1.2G").
  • head -20 keeps just the top 20 results.

Variations:

du -ah . | sort -rh | head -20        # current directory instead of the whole filesystem
du -sh */                             # top-level directories only, not every file
find /var -size +500k -exec ls -lh {} \;   # only files over a size threshold

Restricting to top-level directories first (du -sh */) is often the faster starting point on a large filesystem: it tells you which subtree to dig into with the full du -ah command, rather than sorting through every individual file up front.

If the culprit turns out to be an old backup archive rather than a single runaway file, checking what's actually inside it before deleting is worth the extra step: tar -tzvf backup.tar.gz lists contents with sizes without extracting anything (see tar). For repeated disk investigations, the interactive tool ncdu (not installed by default; apt install ncdu) gives the same information as the du command above but lets you navigate the tree interactively instead of re-running the command with a different path each time.