Hosting Operations

Find Directories Burning Inodes

Inode usage is high and you need to locate which directories contain the most regular files.

Command

find /lab/disk-inode-cleanup/var/cache/app -xdev -type f -printf '%h\n' | sort | uniq -c | sort -nr | head

What changed

Nothing changes. The command counts files by parent directory.

Danger

safe

When to use it

Use when df -ih is high and large-file searches do not explain write failures.

When not to use it

Do not assume the largest count is safe to delete; identify the application owner and cache semantics first.

Undo or recovery

No undo needed because this is a read-only file count.

Expected output

A count of regular files grouped by parent directory, sorted highest first.

demo script

Disposable terminal steps

df -ih /lab/disk-inode-cleanup
find /lab/disk-inode-cleanup/var/cache/app -xdev -type f -printf '%h\n' | sort | uniq -c | sort -nr | head

Terminal transcript Demo JSON

simulated output

What it looks like

disposable vessel

::fixture-ready::
$ df -ih /lab/disk-inode-cleanup
Filesystem     Inodes IUsed IFree IUse% Mounted on
/dev/vda1        512K  499K   13K   98% /lab/disk-inode-cleanup
tmpfs            245K    14  245K    1% /run
::exit-code::0
$ find /lab/disk-inode-cleanup/var/cache/app -xdev -type f -printf '%h\n' | sort | uniq -c | sort -nr | head
    120 /lab/disk-inode-cleanup/var/cache/app/shards/a
     40 /lab/disk-inode-cleanup/var/cache/app/shards/b
::exit-code::0

YouTube Short

Find inode hotspots.

When inodes are the problem, count files by directory. The biggest byte consumer may not be the inode culprit.

experiments

A/B tests to run

Metric: save_rate

A: Inodes are file count pressure.

B: Find the directory burning inodes.