Hosting Operations

Keep du on One Filesystem

You need a directory-level size view without crossing filesystem boundaries during disk triage.

Command

du -xh --max-depth=1 /lab/disk-inode-cleanup/var 2>/dev/null | sort -h

What changed

Nothing changes. The command measures apparent usage and keeps the scan on one filesystem.

Danger

safe

When to use it

Use after df identifies the pressured mount and you need a scoped directory ranking.

When not to use it

Do not run broad recursive scans from / during a busy incident unless the extra I/O is acceptable.

Undo or recovery

No undo needed because du only reads metadata and file sizes.

Expected output

A human-readable size list for direct children under the selected var directory.

demo script

Disposable terminal steps

df -h /lab/disk-inode-cleanup
du -xh --max-depth=1 /lab/disk-inode-cleanup/var 2>/dev/null | sort -h

Terminal transcript Demo JSON

simulated output

What it looks like

disposable vessel

::fixture-ready::
$ df -h /lab/disk-inode-cleanup
Filesystem      Size  Used Avail Use% Mounted on
/dev/vda1        25G   24G  680M  98% /lab/disk-inode-cleanup
tmpfs           982M   12M  970M   2% /run
::exit-code::0
$ du -xh --max-depth=1 /lab/disk-inode-cleanup/var 2>/dev/null | sort -h
8.0K	/lab/disk-inode-cleanup/var/log
8.0K	/lab/disk-inode-cleanup/var/tmp
664K	/lab/disk-inode-cleanup/var/cache
684K	/lab/disk-inode-cleanup/var
::exit-code::0

YouTube Short

Scope your disk scan.

After df names the full mount, use du with xdev behavior and a max depth so the scan stays targeted.

experiments

A/B tests to run

Metric: completion_rate

A: Do not scan the whole machine.

B: Keep du on the pressured filesystem.