command: du

[gcawood]

Finding out how much space your files are using in Linux is a bit of a chore compared to the Windows world. (right click, properties) However, Linux comes with a handy tool `du` that will list how much disk space your files occupy.

USAGE
The most standard usage is the give `du` either a filename

du -ha filename

Another very popular way to use `du` is to find out how large a given directory is.
The `-s` displays the total space that is used for each directory, but not subdirectories.

du -sh directoryname

To find out how much space multiple directories are using. For instance, after taking a backup.

du -sh *

COMMON FLAGS
-h (make it human readable)
-s (total space in a directory, but do not display the space that subdirectories use.
-a (show how much space a file is using)


Gotcha's
`DU` isn't perfect, and if you are a admin long enough, you will find that it doesn't always give you accurate information. Especially if you work with backup archives where one file is hardlinked to another. "Hard Linking" allows one file to exist in multiple locations in a Linux filesystem, but does not duplicate the space requirement. This is a very hand benefit of Linux, especially when you considering backing up data wherever files do not change very often. Using hard links you can have two complete data-sets, but only use the space of one!

Gotcha Example
If you do a `du -sh directory1` and a `du -sh directory2`, where directory2 is a hard link of directory1, it will calculate the space of directory1, twice. Giving you twice as much disk space in use than what is actually being used. The way to get around this is to compare multi-linked inodes by.

Here is a script that I found found handy to use that will calculate the correct space for a directory regardless of whether or not it contains hard links or not.

#!/usr/bin/perl -w

use strict;

my %hash; # Keep track of multi-linked inodes
my $total_blocks = 0;

my @todo = @ARGV;
push(@todo, '.') unless @todo;

foreach my $fn (@todo) {
    my($inode,$nlinks,$blocks) = (lstat($fn))[1,3,12];
    if (-d _) {
	opendir(DP, $fn) or die $!;
	push(@todo, map("$fn/$_", grep(!/^\.\.?$/, readdir(DP))));
	closedir DP;
    } elsif ($nlinks > 1) {
	if (!defined($hash{$inode})) {
	    $hash{$inode} = $nlinks - 1;
	    next;
	}
	next if --$hash{$inode};
    }
    $total_blocks += $blocks;
}

print "$total_blocks blocks\n";
print int($total_blocks / 2 + .5), "K\n"; # Assumes 512-byte blocks

You can find more information about this script, it's original author, and the DU and hard links problem at http://lists.samba.org/archive/rsync...ne/009882.html

[MustangV10]

Quite a helpful command, haven't heard of it before.

Just tried it and it just comes up with a list of file names :S

[carbon333]

-c is also useful switch. It returns total amount of data in every file.

[lanlanlangzi]

Thanks.

[dale]

I use this from time to time:

cd /; du -bS|sort -n

Produces a quick and dirty overview of big files.

[Stefano Messicano]

Another example If you want to show the disk usage of a directory you simply put that directory name as the last argument:
du -h -s movies

[sahabcse]

I am always using the following option with du

c s h

c - Display a grand total
s - Display an entry for each specified file
h - "Human-readable" output. Use unit suffixes: Byte, Kilobyte,
Megabyte, Gigabyte, Terabyte and Petabyte

Ex)
vboxesktop User$$ pwd
/Users/users/Desktop
vboxesktop User$$ du -csh ./*
64K ./1.png
48K ./10.png
36K ./11.png
28K ./12.png
804K ./13.png
484K ./14.png
500K ./15.png
524K ./16.png
572K ./17.png
492K ./18.png
476K ./19.png
112K ./2.png
748K ./20.png
290M ./21
92K ./3.png
84K ./4.png
108K ./5.png
116K ./6.png
124K ./7.png
108K ./8.png
152K ./9.png
1.1M ./Images
15G ./Desktop
4.0K ./Network Recording Player
15G total