3 answers
You are accessing this answer with a direct link, so it's being shown above all other answers regardless of its score. You can return to the normal view.
If you don't mind installing something I recommend ncdu. It sorts the directories and files by size, you can exclude other mounts, navigate the tree directly and even delete files and folders as you go.
ncdu 1.19 ~ Use the arrow keys to navigate, press ? for help
--- / --------------------------------------------------------------------------
14.4 GiB [###########] /home
8.3 GiB [###### ] /var
5.9 GiB [#### ] /usr
1.6 GiB [# ] /opt
180.6 MiB [ ] /boot
77.8 MiB [ ] /root
12.2 MiB [ ] /etc
6.6 MiB [ ] /tmp
3.6 MiB [ ] core
4.0 KiB [ ] /snap
0.0 B [ ] /media
0.0 B [ ] /mnt
@ 0.0 B [ ] initrd.img.old
@ 0.0 B [ ] initrd.img
@ 0.0 B [ ] vmlinuz.old
*Total disk usage: 30.5 GiB Apparent size: 29.2 GiB Items: 667060
There should be a package for it by default in every distribution.
0 comment threads
If you want to go with command-line tools, the first step might be to run a
du -h --max-depth=1 | sort -h -k 1 -r
in your root directory. This will give you a list of the sub-directories sorted by gross usage. You may then proceed by applying the same command inside the sub-directory identified as the "heaviest", to ultimately find the files that eat up all your storage space.
Note that for a "complete picture" you may need to do this with sudo privileges to access directories that your regular user may not enter. Also keep in mind that there are limitations to the accuracy of disk usage reported by du, as explained e.g. in the Wiki page.
If you want to go for a graphical tool, there is the QDirStat project which will represent the individual files as tiles (with sizes proportional to their ... size).
0 comment threads
I agree with other answers:
- Normal TUI way is
ncdu - Normal CLI way is
du - Normal GUI way is Baobab aka "Gnome Disk Usage Analyzer" and the like
But just for fun, you can build a pipeline such as the one below, which finds all files, then uses stat to query their size and parallel to parallelize it, and finally sort and tail to print the best one.
$ fd | pv | parallel 'echo $(stat -c %s {}) {}' | sort -h | tail -n 10
The performance of this is iffy, so I added pv for some progress info. pv doesn't know how many files there are until fd is done enumerating them, so it won't give % progress, only how much is done. You can convert byte sizes to human size with cut and numfmt.
This does give you more control over the specifics, such as if you want to look at only a certain subset of files. You can do that with du as well, but this way you get to use more familiar generic tools rather than dus specialized syntax.
0 comment threads
1 comment thread