Common Linux Commands

This page collects some Linux commands you will use often on VIS Lab servers.

1. Check Where You Are

Right after logging in to a server, it is a good idea to confirm your user, server, and current directory.

Command	What it does
`whoami`	Shows your login username.
`hostname`	Shows the server hostname.
`pwd`	Shows the current directory.
`ls`	Lists files in the directory.
`ls -lh`	Shows file sizes.
`ls -la`	Also shows hidden files.
`history`	Shows recently used commands.

A common check after login:

whoami
hostname
pwd
ls -lh

2. Move Between Directories

Command	What it does
`cd folder_name`	Enter a directory.
`cd ..`	Go up one level.
`cd ~`	Go back to your home directory.
`cd -`	Go back to the previous directory.
`cd /absolute/path`	Enter an absolute path.

If a path has spaces, wrap it in quotes:

cd "folder with spaces"

3. Basic File and Folder Operations

Command	What it does
`mkdir folder_name`	Create a folder.
`mkdir -p path/to/folder`	Create nested folders; missing parent folders are created too.
`touch file.txt`	Create an empty file, or update its timestamp.
`cp source.txt target.txt`	Copy a file.
`cp -r source_folder target_folder`	Copy a folder.
`mv old_path new_path`	Move a file or folder.
`mv old_name.txt new_name.txt`	Rename a file.
`rm file.txt`	Delete a file.
`rm -r folder_name`	Delete a folder.

rm usually cannot be undone

rm does not move files to the trash. Before running it, at least run:

pwd
ls -lh

Confirm which directory you are in and what you are about to delete.

Be especially careful with commands like:

rm -rf folder_name
rm -rf *

On a shared server, do not run rm -rf if you do not understand the path, and do not delete other people's directories.

4. View File Contents

Command	What it does
`cat file.txt`	Print the whole file at once. Good for small files.
`less file.txt`	View a file page by page. Good for large files; press `q` to quit.
`head file.txt`	Show the beginning of a file.
`head -n 20 file.txt`	Show the first 20 lines.
`tail file.txt`	Show the end of a file.
`tail -n 50 file.txt`	Show the last 50 lines.
`tail -f log.txt`	Watch log updates in real time.

When training models, checking logs is very common:

tail -f train.log

Press Ctrl + C to leave tail -f.

5. Search Files and Text

Find Python files under the current directory:

find . -name "*.py"

Find files whose names contain config:

find . -iname "*config*"

Search text under the current directory:

grep -R "learning_rate" .

Search only in Python files:

grep -R "learning_rate" --include="*.py" .

If rg is installed on the server, you can also use ripgrep. It is usually faster:

rg "learning_rate"

6. Edit Text Files

Common command-line text editors on servers include nano, vim, and emacs. If you are not familiar with terminal editors at all, I recommend using VS Code Remote SSH first.

For beginners, nano is the easiest one to start with:

nano file.txt

In nano:

Action	Shortcut
Save	`Ctrl + O`, then press Enter
Exit	`Ctrl + X`
Cancel current action	`Ctrl + C`

7. Check Disk and Directory Usage

Check server disk partition usage:

df -h

Check the size of the current directory:

du -sh .

Check the size of each item in the current directory:

du -sh *

Show current directory contents sorted by size:

du -sh -- ./* ./.??* 2>/dev/null | sort -hr

8. Check CPU, Memory, and Processes

Command	What it does
`lscpu`	Shows CPU model and core information.
`top`	Shows current CPU, memory, and process usage.
`htop`	A more visual CPU, memory, and process view; it may not be installed on every server.
`free -h`	Shows overall memory usage.
`ps -u $USER`	Shows processes currently running under your user.
`ps aux`	Shows the process list on the system.

Search by process name:

ps aux | grep python

View details for one process:

ps -fp PID

Replace PID with the actual process ID.

Stop a process you started:

kill PID

If normal kill does not work, and you have confirmed the process is yours and should be stopped, use:

kill -9 PID

9. Check GPU Usage

Deep learning and graphics computing tasks usually need GPUs. Before starting training, first check whether the GPU is free, so you do not accidentally take resources someone else is using.

Check GPU status:

nvidia-smi

Refresh once per second:

watch -n 1 nvidia-smi

For more rules about shared GPU usage, see GPU / Disk / Memory Guidelines.

Command	What it does
`who`	Shows users currently logged in to the server.
`w`	Shows logged-in users and what they are running.
`last`	Shows recent login records.

These commands help you check whether other people are using the server.

11. Compress and Extract Files

Extract a .zip file:

unzip file.zip

Create a .zip archive:

zip -r archive.zip folder_name

Extract a .tar.gz file:

tar -xzf archive.tar.gz

Create a .tar.gz archive:

tar -czf archive.tar.gz folder_name

See what is inside a .tar.gz file:

tar -tzf archive.tar.gz | head

12. Permission Commands

View file permissions:

ls -l file.txt

Make a script executable:

chmod +x run.sh

Fix SSH private key permissions:

chmod 600 ~/.ssh/id_ed25519_vis

Fix .ssh directory and authorized_keys permissions:

chmod 700 ~/.ssh
chmod 600 ~/.ssh/authorized_keys

Do not casually use chmod -R 777

chmod -R 777 makes files under a directory readable, writable, and executable by everyone. On a shared server, this is usually not what you want. If you run into permission problems, first check who owns the file, where you are, and which permissions really need to be opened.

13. Useful Command-line Tips

Command or symbol	What it does
`Ctrl + C`	Interrupt the command currently running in the foreground.
`Ctrl + L`	Clear the screen.
`Tab`	Auto-complete a command or path.
`Up` / `Down`	View the previous or next command.
`command > out.txt`	Write output to a file, overwriting old content.
`command >> out.txt`	Append output to the end of a file.
`command 2> err.txt`	Write error messages to a file.
`command --help`	Show command help.
`man command`	Show the command manual; press `q` to quit.

Show training output on the screen and write it to a log file at the same time:

python train.py 2>&1 | tee train.log

Append to an existing log:

python train.py 2>&1 | tee -a train.log