Useful Linux commands
The one-stop shop for all your Linux command needs. Most commands listed will work on our machines, though that can vary based on installed packages and/or operating systems.
For almost every command, you can use the --help argument, <command> --help
to get detailed information about the command as well as a list of its arguments.
Contents
File and Directory Management
ls
Lists all files and directories
ls -a
- List all files, including hidden ones like .ssh, .bashrc, and so on.ls -l
- Include extra file information like permissions, owner, groups, modified date, and more.ls -s
- List size of files in blocks.ls -S
- List by size, largest first.
cd
Used to change the directory.
cd
- When no path is specified, you will move to the user's home directory. As root, you will move to/root/
. As your user, you will move to/eccs/home/<username>/
.cd /path/to/file
- When changing directories, if you're not in the immediate parent directory, you need to specify the full path. For example, if my current working directory is/etc
, I cancd jupyterhub
since it's within/etc
. However, if my current working directory is/bin
, I would have to specify the full path for/etc/jupyterhub
.cd .
andcd ..
- The . and .. are used for adjacent movement.cd .
"moves" you into the current directory, so basically no movement.cd ..
moves you backwards one directory.cd -
- The dash specification takes you to the previous working directory regardless of adjacency.
pwd
Displays the current working directory. This can be usual on certain machines that don't display the full path of your current working directory, like Whedon.
root@w0 plugins # pwd /etc/nagios/plugins
In the above example, my current working directory is /etc/nagios/plugins
, but the prompt only lists it at nagios.
mkdir
Used to create directories.
- You can use
mkdir directory_name
to create it in your current working directory. - Additionally, you may specify the full path to create the directory somewhere else
mkdir /path/to/directory_name
.
rmdir
Used to remove empty directories.
- You can ONLY remove empty directories. You cannot specify recursive or force.
mv
Used to move and rename files.
- This command uses a
mv /path/to/file /path/to/destination
syntax. It follows the same rules ascd
in that you need to specify the full path unless moving to the current working directory or an adjacent directory. - To rename a file, you can "move" it into a new one.
mv old.txt new.txt
will put the contents of old.txt into the newly created new.txt. This action will delete old.txt. NOTE: This WILL overwrite the file if you move it into a preexisting one. mv -i
- If you're worried about overwriting files, you can use the -i specification. This will prompt you for confirmation before overwriting.
rm
Used to remove files and/or directories. Unlike rmdir
, you can use rm
to remove directories. Even ones that aren't empty.
rm -d
- Essentially the same asrmdir
; it can only remove empty directories.rm -r
- Recursively removes files and directories from a directory, including itself.rm -rf
- The same process asrm -r
, except it bypasses any locks or protections on files and deletes them. Only use this command if you are certain you understand what you're deleting, especially if you're root.- The
*
isis a powerful wildcard in Linux that allows you to match multiple files efficiently when using commands likerm
. For example, take a look at this directory:
tdjones22@bowie:~/directory$ ls 123.txt 131.txt 14.txt 1file.txt 208.txt 263.txt 62.txt code.py code.rb config.confLet's say you want to remove every file that starts with the number 1. You can use
rm *1
to remove any file matched with the first character being 1.
Files left after the command:208.txt 263.txt 62.txt code.py code.rb config.confLet's say you want to remove files with a specific extension. You can use
rm *.txt
to remove every file with the .txt extension.
Files left after the command:code.py code.rb config.conf
cp
Used to copy files and/or directories. Similar to rm
in that you need to specify recursive when copying a non-empty directory. It follows a cp /path/to/file /path/to/destination
syntax.
cp -r
- Recursively copies a directory; the only way, usingcp
, to copy a directory that isn't empty.
find
A tool to find files using many criteria including name, type, contents of file, and more. A common syntax for this command is find /starting/directory -specification "name_of_file.txt"
.
find / -name "name.txt"
- A typical usecase forfind
is searching by file name. If I've lost track of a file namedfound_file.txt
, I can search the entire machine for it using this command:
tdjones22@bowie:~$ find / -name "found_file.txt" 2>/dev/null /earlhamcs/eccs/users/tdjones22/directory/found_file.txtThis command will usually take some time to execute fully, so don't be alarmed when it hangs for a few seconds.
The starting directory was specified as/
which is just the parent directory, meaning it searches the entire file system. The search type was specified as-name
, which looks for matching named files. You can additionally add2>/dev/null
to silence all of the permission errors you'll get when searching all of the machine's directories.
find / -name "*.txt" -exec grep -iIl "contents to search for" {} + 2>/dev/null
- If you know a portion of the contents of a file, but you don't know the name, you can usefind
in tandem withgrep
to find it. Though, it's far easier and more convenient to only usegrep
.
Let's say I have a missing text file, but I know it contains "solar power", I can find it like so:
tdjones22@bowie:~$ find / -name "*.txt" -exec grep -RiIl "solar power" {} + 2>/dev/null /eccs/home/tdjones22/test_find.txtLike the previous command, this will take a hot minute to complete.
The starting directory was specified as/
which is just the parent directory, meaning it searches the entire file system. We're looking for a file by its name specified as"*.txt"
, which looks for any files with the .txt extension. The-exec grep -iIl "solar power" {} +
section is what searches through the file's contents. The grep arguments specify the following:-l
- ignore case,-I
- ignore binary files, and-l
- display file name. You can additionally add2>/dev/null
to silence all of the permission errors you'll get when searching all of the machine's directories.
File Permissions and Ownership
chmod
Used to add, remove, or change file permissions. You normally specify permissions using an octal format, but you may also use symbolic as well. Here are some common chmod commands; full access for a user implies read, write, and execute ability:
chmod 600
|rw-------
- The owner can read/write, but there is no access for others. Some use cases are private files and SSH keys.chmod 644
|rw--w-r--
- The owner can read/write, but others can only read. Useful for config files and public web files.chmod 660
|rw-rw----
- The owner and group can read/write, but there is no access for others. Useful for files used specifically in group collaboration.chmod 664
|rw-rw-r--
- The owner and group can read/write, but others can only read. Useful for files used specifically in group collaboration.chmod 700
|rwx------
- The owner has full access, but there is no access for others. Useful for private scripts (since you need the ability to execute a script) and sensitive directories.chmod 755
- |rwxr-xr-x
The owner has full access, but others can only read and execute. Useful for certain executables and web directories.chmod 775
|rwxrwxr-x
- The owner and group have full access, but others can only read and execute. Useful for group-managed directories.chmod 777
|rwxrwxrwx
- Everyone, and I mean everyone, has full access. So...don't use this unless absolutely necessary.
chown
Used to change which user, group, or both own files and directories.
chown user file
- Change the owner offile
touser
.chown user:group file
- Change both owner and group offile
.chown :group file
- Change only group offile
.chown -R user:group directory/
- Recursively change owner and group for all files indirectory/
.
chgrp
Used to change only the owner of both files and directories.
chgrp group file
- Change the group offile
togroup
.chgrp group directory/
- Change the group ofdirectory/
.chgrp group file1 file2 file3
- Change the group of multiple files.chgrp -R group directory/
- Recursively change group for all files, including the directory itself, indirectory/
.
stat
Used to view various stats about files and directories such as size, blocks, type, inode, UID/GID, last access, creation date, and more. Let's see what happens when we run stat
on the file test_find.txt
:
tdjones22@bowie:~$ stat test_find.txt File: test_find.txt Size: 27 Blocks: 8 IO Block: 4096 regular file Device: fe04h/65028d Inode: 25452759 Links: 1 Access: (0644/-rw-r--r--) Uid: (11035/tdjones22) Gid: (10000/sharedusers) Access: 2025-03-28 13:42:16.638585831 -0400 Modify: 2025-03-28 11:31:47.781444104 -0400 Change: 2025-03-28 11:31:47.781444104 -0400 Birth: 2025-03-28 11:31:47.781444104 -0400
Let's talk about the output:
Size: 27
- The file is 27 bytes.Blocks: 8
- The file occupies 8 blocks on the disk.regular file
- This is a "regular file", not a directory, symlink, etc.Device: fe04h/65028d
- The device ID of where the file is stored.Inode: 25452759
- The inode number assigned to the file. Every file and directory has a unique inode number.Links: 1
- The number of hard links to this file.(0644/-rw-r--r--)
- Permissions in octal and symbolic format.Uid: (11035/tdjones22) Gid: (10000/sharedusers)
- The user ID (Uid) of the owner and the group ID (Gid).Access: 2025-03-28 13:42:16.638585831 -0400
- The last time the file was read.Modify: 2025-03-28 11:31:47.781444104 -0400
- The last time the file's contents were modified.Change: 2025-03-28 11:31:47.781444104 -0400
- The last time metadata (permissions, ownership, etc) was changed.Birth: 2025-03-28 11:31:47.781444104 -0400
- When the file was first created.
Running this command on a directory displays the same format of output.
Process Management
ps
Used to display information about running processes. There are two common argument options when using ps
: Unix options, which need to be preceded by a dash, and BSD options, which can be grouped and do not need a dash.
ps
is a rather deep command with many specifications and shorthand syntax that is not necessarily intuitive. Here is the link to the official Bash documentation for PS1, which goes heavily into detail about the command with useful quick links at the top of the page.
Here are some common uses:
ps
- Shows only the processes running in your current terminal session.
This is typical output for running
ps
alone:PID TTY TIME CMD 1980517 pts/0 00:00:00 bash 1980970 pts/0 00:00:00 ps
PID
- The Process ID (PID) is the unique number that identifies a process.TTY
- TTY indicates the computer terminal that executed the process.TIME
- The total CPU time the process has used since it started.CMD
- The full command used to initiate the process. Includes paths and arguments.
ps -e/ps -A
- Lists all of the running processes on the machine using the standard syntax.ps -ef
- Lists all of the running processes using the full format. Let's look at what it adds on top of the standardps
:
UID
- The User ID (UID), usually a username, that owns the process.PPID
- The Parent Process ID (PPID).C
- The CPU usage in percentage over the process's runtime.STIME
- The time inHOUR:MINUTE
format that the process started.
ps -u user
- List all of the running processes for a specified user.ps -eH
- List all of the running processes in a hierarchical view. It visually indents processes in a human-readable syntax.
920 ? 00:00:29 sshd 1980491 ? 00:00:00 sshd 1980516 ? 00:00:00 sshd 1980517 pts/0 00:00:00 bash 1983776 pts/0 00:00:00 psThis is the process hierarchy of my bash shell (1980517), via ssh, running the
ps -eH
command.
ps aux
- List all processes (a) for all users (u) using the wide syntax (x).
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND tdjones+ 1984316 0.0 0.0 9984 3488 pts/0 R+ 14:17 0:00 ps auxLet's talk about the wide syntax columns:
%CPU
- CPU usage.%MEM
- Memory usage.VSZ
- Virtual memory size in KB. It includes all code, data, and shared libraries.RSS
- Resident Set Size (RSS) is the actual physical memory used in KB.STAT
- The state of the command.
ps -o
- The-o
argument allows you to specify the format of the output. Here is a command with all of the possible specifications:
ps -o user,pid,ppid,%cpu,%mem,vsz,rss,tty,stat,start,time,cmdYou can specify these in any order.
top
Shows an updating list of currently running processes, along with their CPU and memory usage. When running top
in your terminal, you'll be prompted with a command line interface (CLI) like so:
Tasks: 276 total, 3 running, 273 sleeping, 0 stopped, 0 zombie %Cpu(s): 25.2 us, 2.1 sy, 0.0 ni, 72.0 id, 0.0 wa, 0.0 hi, 0.7 si, 0.0 st MiB Mem : 193403.2 total, 81554.2 free, 2989.2 used, 108859.8 buff/cache MiB Swap: 66756.0 total, 66213.6 free, 542.4 used. 189014.0 avail Mem PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1610433 qkalaya+ 20 0 13836 5796 3088 R 94.1 0.0 15822:10 python3.10 1610629 qkalaya+ 20 0 13832 5808 3088 R 94.1 0.0 15821:10 python3.10 2508275 mgcithi+ 20 0 11.2g 78884 44700 S 5.9 0.0 1:19.80 node 2553235 tdjones+ 20 0 10520 3952 3364 R 5.9 0.0 0:00.02 top
Here are some handy controls once you're in the CLI:
q
- Quits and exits the CLI.Ctrl + C
also works for exiting, just not as gracefully, perhaps.k
- Kills the process identified by an input process ID (PID). A more interactive and readable alternative to usingkill
r
- 'Renice' a process; change the processes's priority. The 'niceness value' of a process determines how often it receives CPU time relative to other processes. The actual value itself ranges from -20 to 19, with -20 being the highest priority and 19 being the lowest.P
- Sort processes by CPU usage. This is the default when openingtop
.M
- Sort processes by memory usage.T
- Sort processes by time running.
Additionally, if you want or need to take a snapshot of top
's data at a single moment, you can run top -b
. You can also save its output to a file using top -b -n 1 > output.txt
. It's important to specify -n 1
, which tells it to capture the batch once; otherwise, it'll loop forever.
kill
Used to kill a single running process using its process ID (PID). kill
has a few options that you'll commonly use. Here are some:
kill -15 PID
- This is the same as runningkill
without any specifications. It gracefully shuts down the process, allowing it to save its current state, close corresponding files, remove temporary files, log its shutdown process, and notify other processes/systems.kill -9 PID
- This really kills a process. It skips all cleanup procedures and forcefully stops the process and anything associated with it.
pkill
Used to kill any processes that match the input name. Unlike kill
, this can terminate multiple processes. Here are some common uses:
pkill process_name
- This will cleanly terminate the specified process.pkill -f process_name
- This allows you to terminate a process based on any matching string within the full command line. For example, if I runpgrep -fa python3
, I can see all of the processes that would be terminated if I swappedpgrep
forpkill
:
tdjones22@bowie:~$ pgrep -fa python3 899 /usr/bin/python3 /usr/bin/fail2ban-server -xf start 1569 /usr/bin/python3 /usr/share/unattended-upgrades/unattended-upgrade-shutdown --wait-for-signal 1610433 /usr/local/bin/python3.10 /eccs/home/qkalaya24/sll.py 1610629 /usr/local/bin/python3.10 /eccs/home/qkalaya24/sll.pyI'm able to kill a process based on it using any version of python3. This is true for anything you see listed in the full process.
pkill -u user process_name
- This will only match processes owned by the specified user.pkill -x process_name
- This kill a process only if it's name matches exactly.pkill -n process_name
- If there are multiple matching processes, it will kill the newest matching one.pkill -o process_name
- If there are multiple matching processes, it will kill the oldest matching one.pkill -9 process_name
- Force quits the process.
nice
Used to adjust how much CPU time a process gets relative to other processes. Every process has a nice value between -20 (highest priority) and 19 (lowest priority). Niceness levels below 0 must be set by root.
nice -n value command
- When runningnice
, it means you're starting the command. For example,nice -n 15 python3 code.py
sets the niceness value and then executes the following command.
renice
Used to adjust how much CPU time an already running process gets relative to other processes. Every process has a nice value between -20 (highest priority) and 19 (lowest priority). Niceness levels below 0 must be set by root. Renice only accepts process IDs (PIDs).
renice 10 -p 1234
- This will change the niceness level of the process correlated with PID 1234 to a level of 10.
User Management
whoami
This command prints the username associated with the effective ID. If I run whoami
as my stand user, this is the output I'd expect:
tdjones22@bowie:~$ whoami tdjones22
Simple and straight to the point.
However, for a fun example of how effective UID (EUID) works vs real UID (RUID), run whoami
as your user with sudo
:
tdjones22@bowie:~$ sudo whoami root
who
Lists current users who are logged in with information regarding their username, the date and time of login, and the IP address they were connected to when logged in.
w
Lists current logged-in users with additional information such as username, from address, login date, idle time, CPU usage, and the last command they used.
id
This command shows your user ID, group ID, and the IDs of any other groups or privileges you have. Running the command with no specification defaults to showing your IDs:
tdjones22@hopper:~$ id uid=11035(tdjones22) gid=10000(sharedusers) groups=10000(sharedusers),1210(cybersecurity),5050(sudoers)
However, you may specify a username and see their IDs as well:
tdjones22@hopper:~$ id charliep uid=10042(charliep) gid=10000(sharedusers) groups=0(wheel),121(software),1007(bccd),116(mothur),122(fieldscience),1208(production),120(staff),170(hip),173(pedagogical),176(cs-department),1030(cvs),5050(sudoers),1108(fieldsci),1109(webdev),1107(dpr),1115(nbgrader),1200(csfaculty),1206(cluster),10000(sharedusers)
passwd
This command allows you to change the password of the specified user. When changing your own password, you will be prompted to enter your old one before you can set the new one:
tdjones22@hopper:~$ passwd tdjones22 Enter login(LDAP) password: New password:
If you or someone you know has lost their password, you can bypass the LDAP authentication by either being logged in as root or using sudo
:
tdjones22@hopper:~$ sudo passwd tdjones22 New password:
su
This command allows you to switch to another user, assuming you have privileges and/or the password. This is a vital command in the System Administration world because it's the most common way to become root:
tdjones22@hopper:~$ sudo su - root@hopper:~#
You'll notice that we're using -
in place of root. Here's what happens if you switch user directly to root:
tdjones22@hopper:~$ sudo su root root@hopper:/cluster/home/tdjones22#
It preserves the current working directory you were in before switching to root, which, in my case, was my personal home directory. cd -
takes you to the directory you last visited, but in the context of sudo su -
, it takes you to root's home directory /root
.
sudo
Short for superuser do, sudo
is one of the most powerful tools in the System Administrator toolbox. When logged into an account with sudo/sudoers access, you can run any command with root privileges. It temporarily changes your effective UID (EUID) to match that of root's EUID, giving you root privileges.
Disk and Filesystem Management
df
Standing for disk free, df
provides information on disk space utilization, specifically for filesystems.
df -h
- Perhaps the most common way to usedf
, this specification converts the bytes into *human* readable syntax: MB, GB, TB, and so on.
tdjones22@bowie:~$ df -h Filesystem Size Used Avail Use% Mounted on udev 95G 0 95G 0% /dev tmpfs 19G 1.6M 19G 1% /run /dev/mapper/virtual0-root 274G 186G 75G 72% / tmpfs 95G 2.1M 95G 1% /dev/shm tmpfs 5.0M 0 5.0M 0% /run/lock /dev/mapper/virtual0-docker 1015G 300G 665G 32% /docker /dev/mapper/virtual0-eccs--lv 4.0T 2.7T 1.2T 70% /earlhamcs /dev/mapper/virtual0-postgres 1007G 457G 499G 48% /postgres tmpfs 19G 0 19G 0% /run/user/11162 tmpfs 19G 0 19G 0% /run/user/11170 tmpfs 19G 0 19G 0% /run/user/11035Let's talk about the output:
- Filesystem - The name of the mounted filesystem.
- Size - The total size of the filesystem.
- Used - The amount of space occupied in the filesystem.
- Avail - The remaining usable space in the filesystem.
- Use% - The percent of the filesystem used.
- Mounted on - The directory the filesystem is mounted on.
df -h file/directory
- You can run df -h on a file or directory to get information about the filesystem it lives in.
tdjones22@bowie:~$ df -h output.txt && echo "-------------" && pwd Filesystem Size Used Avail Use% Mounted on /dev/mapper/virtual0-eccs--lv 4.0T 2.7T 1.2T 70% /earlhamcs ------------- /eccs/home/tdjones22
du
Standing for disk usage, du
provides information on the amount of disk usage for files and directories.
du -h
- Perhaps the most common way to usedu
, this specification converts the bytes into *human* readable syntax: MB, GB, TB, and so on.du --max-depth=<number>
- When runningdu
, it displays disk usage for every directory recursively, meaning if your current working directory is somewhere higher up in the hierarchy, you'll get a huge slurry of output.--max-depth=<number>
defines how many recursive steps the command will take:
tdjones22@bowie:~/CS325$ du -h --max-depth=1 8.0K ./.vscode 30M ./Lab2 5.7G ./head_node_backup 5.8G .This shows the disk usage for the base-level directories in your current working directory. NOTE:
du -h -d 0
is the same asdu -h --max-depth=0
. I prefer the additional context of--max-depth=0
.You can also use the depth specifier to see the disk usage for only your current working directory:
tdjones22@bowie:~/CS325$ du -h --max-depth=0 5.8G .
du -a
- By default,du
does not display disk usage of files unless the file is specifieddu file
. However, the -a argument displays disk usage for all, including directories and files.
tdjones22@bowie:~/CS325$ du -h -a --max-depth=1 4.0K ./web_check_v2.txt 4.0K ./artificial_load.md 48K ./artificial_load.pdf 4.0K ./jumbo_frames.md 4.0K ./loop.sh 8.0K ./.vscode 4.0K ./website_check.sh 4.0K ./config_manifest.md 30M ./Lab2 5.7G ./head_node_backup 28K ./jumbo_frames.pdf 8.0K ./final_study_guide.md 40K ./config_manifest.pdf 4.0K ./test.sh 4.0K ./website_check_v2.sh 5.8G .
mount
Used to mount filesystems. This works on filesystems being sent through Network File System (NFS) as well as filesystems on mounted disks (HDDs, SSDs, USB Flash Drives, Memory Cards, etc).
mount -a
- The -a specification tells your machine to mount all available filesystems listed in/etc/fstab
. This is useful for system administration because it's common to mount multiple filesystems being exported from multiple machines.mount /path/to/filesystem
- Specifying the filesystem you want to mount means you will only mount that filesystem. This is the command you'll want to use when mounting a filesystem from a device.
In CS325 (Systems Engineering and Administration), we set up a cluster using Raspberry Pi 5's. The Pi's have two main methods of storage, a micro SD card and a mini NVMe slot. The SD card we were using was faulty, resulting in it crashing and us having to wipe the head node three times. However, when it bricked the final time, I was able to take the SD card, pop it into a machine in Lovelace, and mount it using
mount /dev/sdc2
. The boot partitionsdc1
had become corrupted, but the root partition with all of our data was salvageable. All of this to say,mount
is an extremely useful tool in the Linux world.
umount
Used to unmount filesystems. This works on filesystems being sent through Network File System (NFS) as well as filesystems on mounted disks (HDDs, SSDs, USB Flash Drives, Memory Cards, etc). It's like the "Eject" option you select on Mac or Windows when removing something like an external hard drive to ensure nothing bad happens.
umount -a
- The -a specification tells your machine to unmount all filesystems listed in/etc/fstab
.umount /path/to/filesystem
- Specifying the filesystem you want to unmount means you will only unmount that filesystem. This is the command you'll want to use when unmounting a filesystem from a device.
fsck
Standing for File System Consistency Check, this is used to check and repair filesystems. NOTE: Do NOT run this command on a mounted filesystem as it can very easily corrupt the filesystem data.
- It starts by reading the superblock of the filesystem, which is a metadata structure that contains vital information such as the size, type, status, and layout.
- After reading through the superblock, it confirms filesystem consistency by checking:
- Inode table - Each file/directory has an inode with metadata.
- Block allocation - Ensuring no two files claim the same data block.
- Directory structures - Ensuring that directory entries point to valid inodes.
- Link counts - Comparing actual hard links versus what the inode metadata says.
- Free block/inode bitmaps - Ensuring they reflect what's actually in use.
- If the command detects no issues, it will report some information about the filesystem and conclude its investigation. If it detects errors, it will prompt you for permission to fix them.
I highly recommend doing more intensive research on fsck
before using it, as it can be destructive if used improperly.
-- Major Overhaul: March 2025 by Tyler Jones [tdjones22] --