I want to estimate the amount of disk space used by a directory using the following command.
du -sh dir_name
which does not calculate the hidden directories. In the man page of du there is no info regarding it. How to calculate the amount of disk space used by the directories including the hidden files.
Actually it does, here is the proof:
mkdir .test
echo "hi" > .test/appo
du -a
4 ./.test/appo
8 ./.test
12 .
The -a option is used to explicitly show which files were counted.
Are you using du *?
This command shows you the summarized size of hidden directories using a regular expression:
du -hs .[^.]*
du -hs .[^.]* * will include both dot files/directories and non-hidden directories together. I don't know if there are cases where this is not correct.
du -hs ..[^.]* .[^.]* *
The correct command is : du -hs $(ls -A)
$ du -hs $(ls -A)
0 test
0 .test
du -hs .* *, as mentioned in another answer, is not correct if you want to list all files and subdirectories (including hidden ones).
Example :
$ touch test
$ touch .test
$ echo *
test
$ echo .* *
. .. .test test
$ du -hs .* *
4,0K .
1,8G ..
Why does du behave like this? Because you use -s that summarize the result and that all files and subdirectories are children of . so du -hs does not list them!
ls -A-based solution, but this time using xargs so as to make sure filenames with spaces are properly handled (see my other answer below)
FYI, for estimating the size occupied by various directories, its much better to use ncdu
You can navigate in the ncurses GUI between various directories and it will show the size of each directories. If I am using du, I would have to execute du command for each directory I want to check for which can be cumbersome. You can sort the directories according to the size occupied too in the ncurses GUI.
du. Aside from a fancier display method, there's no obvious indication of any functional difference.
Commented
Oct 4, 2011 at 10:15
d binding to delete directories, so you can clean up directly from the interface.
Commented
Mar 17, 2017 at 10:06
du -ahd1 | sort -hr | head -10
following is the description of -d option in du --help:
-d, --max-depth=N print the total for a directory (or file, with --all) only if it is N or fewer levels below the command line argument; --max-depth=0 is the same as --summarize
It does and it does not. Example:
In the home directory: (only one user exist)
du -sh /home/*
2.6G /home/user
in the user directory: (huge difference between the sums)
du -sh *
61M bin
2.0M dump-20130124104823.tar.gz
651M public_html
472K twitter-2.0.0.gem
11M wkhtmltopdf-0.11.0_rc1-static-amd64.tar.bz2
and the reason is:
du -sh /home/user/.rvm/
1.9G /home/user/.rvm/
du will calculate hidden directories while descending into subdirectories, but in the current directory the * simply does not match to .directory_name pattern so the current directory hidden elementes will be omitted.
It took me some time to figure out, and as shadyabhi recommends it would have been obvious if i had used ncdu.
A solution using ls -A to list files and xargs to manipulate output:
ls -A | xargs -I {} du -sh {}
This solution makes sure that filenames with spaces are correctly handled.
-s instead of -sh to ensure "560K" doesn't sort above "2.0G".
You can use the following command:
find -maxdepth 1 -exec du -sh "{}" \; | sort -h
This gives you:
It also sorts the output to make it easy to see what is the largest.
I also made this an alias in my ~/.bash_aliases file.
alias big='find -maxdepth 1 -exec du -sh "{}" \; | sort -h'
alias sbig='sudo find -maxdepth 1 -exec du -sh "{}" \; | sort -h'
Now I can just execute big to find the biggest directories in a directory and sbig when I need sudo permissions.
This command will help you to check the disk usage, go to your directory and execute the following:
du -sch .[!.]* * | sort -h
Here is a demo of how to calculate disk usage of a particular directory.
I have created a directory called Du_sh and created two files inside of it (one hidden [10 MB] and one normal [20 MB]).
nikhil@debian:~$ mkdir Du_sh
nikhil@debian:~$ cd Du_sh/
nikhil@debian:~/Du_sh$ dd if=/dev/zero of=.10MB bs=1024 count=10240
10240+0 records in
10240+0 records out
10485760 bytes (10 MB, 10 MiB) copied, 0.0299941 s, 350 MB/s
nikhil@debian:~/Du_sh$
nikhil@debian:~/Du_sh$ dd if=/dev/zero of=20MB bs=2048 count=10240
10240+0 records in
10240+0 records out
20971520 bytes (21 MB, 20 MiB) copied, 0.0342258 s, 613 MB/s
nikhil@debian:~/Du_sh$ du -sch .[!.]* *
10M .10MB
20M 20MB
30M total
nikhil@debian:~/Du_sh$ du -sch .[!.]* * | sort -h
10M .10MB
20M 20MB
30M total
nikhil@debian:~/Du_sh$
The correct command is:
du -hs .* *
From man pages:
-h, --human-readable
print sizes in human readable format (e.g., 1K 234M 2G)
-s, --summarize
display only a total for each argument
You can read more about du command here.
Most things I tried (including other answers) didn't work in one way or another, so here is my (currently) final solution for using du:
ls -A | while read file; do du -sh "$file"& done | sort -h;
ls -A All files including hidden, without . and ..while read file; do du -sh "$file"& done Run the du command on each entry from ls -A in parallel. Had no problem yet regarding too much parallelism, but you can either replace & by ; making it synchronous or add [ $( jobs | wc -l ) -ge $( nproc ) ] && wait before donesort -h Sort by sizeWhile I'm at it: I have a function named dus added to my profile to keep this function at hand:
dus () { ls -A | while read file; do du -sh "$file"& done | sort -h; }
The above answer by Guillaume Missonnier, du -hs $(ls -A), does not work if you want sizes of files in a directory other than pwd e.g., du -hs $(ls -A /other/dir/) gives du: cannot access error.
This happens because the output of ls command is not valid under pwd.
I figured a simple way to fix this, first cd to that dir, execute the command and then cd back to pwd. Assume you are in a non-home directory and would like to get sizes of files in home directory (different from pwd). In that case, use:
cd ~; du -sh $(ls -A) | sort -h; cd -
A better way would be to create an alias/function to take the /other/dir as argument, so it will be easy to call as: myfunc /other/dir.
du accepts paths. With this command you can check the disk usage of a different path:
du -sh /path/to/directory/{.[^.],}*
If you want the result sorted (bigger files/dirs first):
du -sh /path/to/directory/{.[^.],}* | sort -hr
I found this comment by Stéphane Chazelas worth a separate answer for zsh users.
On zsh, use:
du -sch *(D) | sort -h
The solutions with .[^.]* * or .[!.]* * do not work on zsh if a folder contains only .* files or * files. It will throw a no matches found error. So you cannot use one command universally.
On bash, it would return the following error for one type but still continue with the other type.
No such file or directory
From my tests the safest approach is this...
du -sm -- * .[^.]*
TIP: Prefer megabytes (-m option).
NOTE: Occasional side effect du: cannot access '.[^.]*': No such file or directory. When there is no hidden files/folders? 🤔
Thanks! 😎
Sortly list by megabytes the 15 largest files/folders...
du -sm -- * .[^.]* | sort -nr | head -15
It seems to me that an approach with ls -A might be safer as this command is suitable for listing files (-A including hiddem ones). However, we have discrepancies in the outputs and problems with single quotes even using xargs to try to handle the situation.
SAMPLES
[eduardolac-pc eduardolac]# du -sm -- * .[^.]* | sort -nr | head -15
553002 Data1
140123 Data0
37718 .PlayOnLinux
17349 .local
13424 .cache
8654 .config
7931 .var
3423 .pyenv
2833 .npm
2051 .stremio-server
2001 .Phoenicis
1896 .wine
1479 .conda
1451 .nvm
1402 .vscode
[eduardolac-pc eduardolac]# ls -A | xargs -I {} du -sm {} | sort -nr | head -15
xargs: unmatched single quote; by default quotes are special to xargs unless you use the -0 option
553002 Data1
140123 Data0
37718 .PlayOnLinux
17349 .local
13416 .cache
8654 .config
2833 .npm
2001 .Phoenicis
1479 .conda
1451 .nvm
962 .goldendict
932 GoldenDict_Glossaries
803 .mozilla
785 .m2
570 .googleearth
[Ref(s).: https://serverfault.com/a/1116989/276753 , https://superuser.com/a/1739605/195840 ]
You may use braces expansion with wildcarding and human readable sorting:
du -hd2 {.,[^.]}* | sort -h
d2 is a depth option
