delete millions of files inside a directory in LINUX, fast, efficient & resource friendly delete tool.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


This application will delete all files inside a directory given from command line as an argument,

The files are not sorted and deleted as the app seek the file.

Please use in caution as the app will delete all files without asking for it.

This app is especially useful when you want to delete millions of files inside a directory and you get the error from rm: "argument list too long" If you suffer from disk empty but you don't have any inodes to use the space, mostly 0 sized millions of files causes this error. Application tested to be faster than rm, rsync and perl alternatives.


Download the code:

git clone

cd to dir:

cd massDelete

compile the code:


or one can simply use gcc:

gcc massDelete.c -o massDelete

to delete files in /var/lib/session:

./massDelete -d /var/lib/session -v

The -v means verbose mode, it is optional.

to delete files in /tmp with 2500 microseconds sleep between each deletion:

./massDelete -d /tmp -v -s 2500

By default, the program removes the empty directory afterwards. To preserve the directory, use the -p option:

./massDelete -d /tmp -p

Recursive Deleting

The program does NOT delete folders recursively. You can use find to achieve it. Let's say you want to delete every file in /opt/some_dir/a and /opt/some_dir/b:

find /opt/some_dir -mindepth 1 -maxdepth 1 -type d -exec massDelete -d {} -v \;


Version 0.5 Now removes the empty directory afterwards. If the users wants to keep the empty directory, has to explicitly provide the -p option.

Version 0.4 Added Average deletes per second

Version 0.3 Added sleep between each delete cycle to drop disk utilization

Version 0.2 Added Ctrl + C handler, now you can see the number of deleted files on hitting CTRL + C

Version 0.1 Initial Release