If you’ve ever ran out of disk space, you know how terribly painful it can be. Applications will stop opening, indexing services crawl. If you have a swap file or partition configured, it will start to shred. Working with a remote server can be even more painful, especially if you don’t have any GUI monitoring services running. Fortunately, we can get everything we need from the command-line using commands like
This article will use the commands
du to analyze disk and partitiona space as well as storage usage by files and directories. These commands are pretty standard on Unix-like operating systems like macOS and Linux.
Neither command requires elevated privileges, unless you are attempting to use
du on a directory that you don’t have permission to. We’ll be working with files and directories that your user account should have access to and none of the commands are destructive in nature.
🐊 Alligator.io recommends ⤵⚛️ Fullstack Advanced React & GraphQL by Wes Bos
df command allows you to get information about the physical disks and partitions of the system you are on. Running the command without any arguments will present you with a list of the disks, their total size, storage used and available, and the mount point on your system:
Will output something like this:
Filesystem 1K-blocks Used Available Use% Mounted on dev 16399220 0 16399220 0% /dev run 16407880 1524 16406356 1% /run /dev/nvme0n1p2 982940416 231329248 701610820 25% / tmpfs 16407880 32584 16375296 1% /dev/shm tmpfs 16407880 0 16407880 0% /sys/fs/cgroup tmpfs 16407880 1698632 14709248 11% /tmp /dev/nvme0n1p1 523248 107668 415580 21% /boot tmpfs 3281576 84 3281492 1% /run/user/1000
The output on your system may look a bit different as there are different implementations of the
df command. The output above is from the GNU version of
The numbers can be a bit intimidating as they can be pretty large and aren’t prettified up with commas or periods. To improve the output, we can add the
--human-readable flag to display the sizes in the power of 1024:
$ df -h
Which changes the output to:
Filesystem Size Used Avail Use% Mounted on dev 16G 0 16G 0% /dev run 16G 1.5M 16G 1% /run /dev/nvme0n1p2 938G 221G 670G 25% / tmpfs 16G 32M 16G 1% /dev/shm tmpfs 16G 0 16G 0% /sys/fs/cgroup tmpfs 16G 1.7G 15G 11% /tmp /dev/nvme0n1p1 511M 106M 406M 21% /boot tmpfs 3.2G 84K 3.2G 1% /run/user/1000
There is also the
--si that will use the powers of 1000 instead of 1024.
Because I’m a human, and prefer to save a few keystrokes whenever possible, I alias
df -h in my shell configuration:
# Disk utility aliases $ alias df='df -h' $ alias du='du -h' # More on du below
If you’re working with a system that has a ton of directories or files, the
inode (or index node) usage may be more important than the block size. If you would prefer to see the
inode usage information, you can pass the
$ df -i
Which presents a similar set of information, but with
inode information instead of block storage information:
Filesystem Inodes IUsed IFree IUse% Mounted on dev 4099805 418 4099387 1% /dev run 4101970 819 4101151 1% /run /dev/nvme0n1p2 62480384 2139558 60340826 4% / tmpfs 4101970 180 4101790 1% /dev/shm tmpfs 4101970 16 4101954 1% /sys/fs/cgroup tmpfs 4101970 220083 3881887 6% /tmp /dev/nvme0n1p1 0 0 0 - /boot tmpfs 4101970 51 4101919 1% /run/user/1000
And assuming you’re a human, you can also include the
-h argument to display the
inode information in a more human-readable format.
du command allows you to see the disk usage information at a more micro level than that of the
df command. Invoking the
du command without any arguments will display the disk usage for the current directory.
If you would prefer to show the disk usage for a directory that isn’t the directory you’re current in, you can simply pass in the path:
$ du $ du . # Same thing, just more explicit $ du ~ # Explicitly analyzing your home directory $ du /home/user # Even more verbose
If you were to run that command from your home directory, you’ll probably be greeted with a ton of output that quite frankly, is pretty intimidating and hard to sift though.
Some truncated results from my own system:
# ... ton of previous output 4712 ./.node-gyp/10.10.0/include 4720 ./.node-gyp/10.10.0 13584 ./.node-gyp 96132 ./.electron 88 ./.ssh 20 ./.parallel 36880032 ./vmware 20 ./Desktop 152 ./Pictures/Webcam 904 ./Pictures 4 ./Music 12 ./.steam 166449800 .
The command lists out every single directory, subdirectory and file and reports the amount of disk space that it’s occupying. The final line of the output is the total disk usage for the directory, at upper most level.
du output defaults to the number of bytes. Both set of arguments,
--human-readable as as well as
--si are available, to help make the output more digestible:
$ du -h
Which outputs the information in the base of 1024:
# ... ton of previous output 4.7M ./.node-gyp/10.10.0/include 4.7M ./.node-gyp/10.10.0 14M ./.node-gyp 94M ./.electron 88K ./.ssh 20K ./.parallel 36G ./vmware 20K ./Desktop 152K ./Pictures/Webcam 904K ./Pictures 4.0K ./Music 12K ./.steam 159G .
And if you’re more interested in the
inode values than the block storage values, you can also use the
Even with the human-readable output, a large directory is going to be pretty noisy since every single directory, subdirectory and file is going to be reported on.
To help clean up the output a bit, we can limit the depth that is output by passing in the
$ du -d 0 ~ # Grand total for the directory $ du -d 1 ~ # Totals for files in the current directory $ du -s ~ # Shorter version of -d 0
Depending on the size of the directory, limiting the max depth to traverse may cause things to appear like they are locked up since there will be a lack of visual output.
du command isn’t limited to a single directory either, if you wanted to find out the disk usage for your home directory as well as
/tmp you can do so quite easily:
$ du -s ~ /tmp
If you’re a macOS user and have tried to get the total size of a handful of files and/or directories from Finder, there’s a good chance you’ve wound up with a bunch of open Information dialogs.
Graphical tools can be quite convenient, but tend to fall short for power user scenarios. Generally speaking, they also tend to not be available on remote systems.
du don’t do much, but they do it well. Getting the size of files and disk usage tends to be a task that comes up when you’re low on space and need to get something resolved quickly.
These commands make it easy to identify problem areas in your system quick and efficiently and are generally available on Linux systems out of the box.