Navigation and Files Cheat Sheet¶

Quick reference for moving around, managing files, reading them, archiving, checking disk usage, and links. Chapters: Your first commands, Working with files, Archives and compression, Filesystems, inodes, and links.

Paths at a glance¶

Path	Meaning
`/`	The root of the whole filesystem
`~`	Your home directory (`/home/alex`)
`~bob`	Another user's home directory
`.`	The current directory
`..`	The parent directory
`-`	(with `cd`) the previous directory
`/etc/hosts`	Absolute path: starts at `/`, works from anywhere
`notes/todo.md`	Relative path: starts at the current directory

Moving around¶

Command	What it does	Example
`pwd`	Print the current directory	`pwd` → `/home/alex/projects`
`cd DIR`	Change directory	`cd /var/log`
`cd`	Go home	`cd`
`cd -`	Go back to the previous directory	`cd -`
`cd ..`	Go up one level	`cd ../..` goes up two
`pushd DIR` / `popd`	Visit a directory, then return to where you were	`pushd /etc/nginx` … `popd`
`ls`	List a directory	`ls ~/Downloads`
`ls -l`	Long format: permissions, owner, size, date	`ls -l /etc/passwd`
`ls -a`	Include hidden "dotfiles"	`ls -a ~`
`ls -lh`	Human-readable sizes (K, M, G)	`ls -lh *.csv`
`ls -lt` / `ls -ltr`	Sort by time, newest first / oldest first	`ls -ltr /var/log` (newest at bottom)
`ls -lS`	Sort by size, largest first	`ls -lS ~/Downloads \| head`
`ls -ld DIR`	Show the directory itself, not its contents	`ls -ld /tmp`
`ls -1`	One name per line (good for scripts)	`ls -1 *.log`
`tree -L 2`	Show a directory tree, 2 levels deep	`tree -L 2 -d ~/projects` (dirs only)

Creating, copying, moving, deleting¶

Command	What it does	Example
`mkdir DIR`	Create a directory	`mkdir reports`
`mkdir -p A/B/C`	Create parents as needed; no error if it exists	`mkdir -p data/2026/10`
`touch FILE`	Create an empty file, or update its timestamp	`touch notes.md`
`cp SRC DST`	Copy a file	`cp config.ini config.ini.bak`
`cp -r SRC DST`	Copy a directory recursively	`cp -r site/ site-old/`
`cp -a SRC DST`	Archive copy: recursive, keeps permissions, times, links	`cp -a /etc/nginx ~/nginx-backup`
`cp -i`	Ask before overwriting	`cp -i new.csv data.csv`
`cp -u`	Copy only if source is newer	`cp -u *.csv backup/`
`mv SRC DST`	Move or rename	`mv draft.md final.md`
`mv -n`	Never overwrite an existing file	`mv -n *.jpg photos/`
`rm FILE`	Delete a file (no trash, no undo)	`rm old.log`
`rm -r DIR`	Delete a directory and everything in it	`rm -r build/`
`rm -I`	Ask once before deleting more than 3 files or recursing	`rm -rI build/`
`rmdir DIR`	Delete an empty directory only	`rmdir empty/`

rm has no undo

There's no trash can on the command line. Check a glob first with ls (ls *.tmp, then rm *.tmp). Never run rm -rf on a path built from a variable that might be empty.

Viewing files¶

Command	What it does	Example
`cat FILE`	Print a whole file	`cat /etc/os-release`
`less FILE`	Page through a file (see keys below)	`less /var/log/syslog`
`head -n N`	First N lines (default 10)	`head -n 5 data.csv`
`tail -n N`	Last N lines	`tail -n 50 app.log`
`tail -f`	Follow a file as it grows	`tail -f /var/log/syslog`
`tail -F`	Follow, and reopen if the file is rotated	`tail -F app.log`
`tail -n +2`	Everything from line 2 (skip a header)	`tail -n +2 data.csv`
`wc -l`	Count lines (`-w` words, `-c` bytes)	`wc -l *.csv`
`file FILE`	Guess the file type from its contents	`file mystery.bin`
`stat FILE`	Size, inode, permissions, timestamps	`stat report.pdf`
`diff -u A B`	Show differences between two files	`diff -u old.conf new.conf`
`cmp A B`	Check if two files are byte-for-byte equal	`cmp a.bin b.bin`
`sha256sum FILE`	Checksum a file	`sha256sum ubuntu.iso`
`realpath PATH`	Full absolute path, symlinks resolved	`realpath ../notes`
`basename` / `dirname`	Last part / directory part of a path	`basename /var/log/syslog` → `syslog`

`less` keys¶

Key	Action	Key	Action
`Space` / `B`	Page down / up	`/text`	Search forward
`G` / `Shift`+`G`	Start / end of file	`?text`	Search backward
`N` / `Shift`+`N`	Next / previous match	`Shift`+`F`	Follow mode, like `tail -f` (`Ctrl`+`C` to stop)
`-S`	Toggle line wrapping	`Q`	Quit

Archives and compression¶

A tarball bundles many files into one .tar file; compression shrinks it. tar does both. Remember the letters: create, extract, list, file, verbose, z gzip, j bzip2, J xz.

Task	Command
Create `.tar.gz`	`tar -czf backup.tar.gz project/`
Create `.tar.xz` (smaller, slower)	`tar -cJf backup.tar.xz project/`
Create `.tar.zst` (fast and small)	`tar --zstd -cf backup.tar.zst project/`
Pick compression from the name	`tar -caf backup.tar.zst project/`
List contents	`tar -tf backup.tar.gz` (add `-v` for details)
Extract here	`tar -xf backup.tar.gz` (compression is auto-detected)
Extract into a directory	`tar -xf backup.tar.gz -C /tmp/restore`
Extract one file	`tar -xf backup.tar.gz project/config.ini`
Print one file to stdout	`tar -xOf backup.tar.gz project/config.ini`
Drop the top-level directory	`tar -xf app-1.2.tar.gz --strip-components=1 -C app/`
Exclude files	`tar --exclude='*.log' -czf src.tar.gz project/`
Copy a directory over SSH	`tar -czf - project/ \| ssh alex@lab 'tar -xzf - -C ~'`

Single-file tool	Compress	Decompress	Read without extracting
gzip (`.gz`)	`gzip big.log` (`-k` keeps original)	`gunzip big.log.gz`	`zcat`, `zless`, `zgrep`
bzip2 (`.bz2`)	`bzip2 big.log`	`bunzip2 big.log.bz2`	`bzcat`
xz (`.xz`)	`xz -T0 big.log` (all CPU cores)	`unxz big.log.xz`	`xzcat`
zstd (`.zst`)	`zstd big.log`	`unzstd big.log.zst`	`zstdcat`
zip (`.zip`)	`zip -r site.zip site/`	`unzip site.zip -d site/`	`unzip -l site.zip`

Look before you extract

Run tar -tf file.tar.gz | head first. If the archive doesn't have a single top-level directory, extract it into a new empty directory with -C, so it doesn't scatter files everywhere.

Disk usage¶

Command	What it does	Example
`df -h`	Free space on each mounted filesystem	`df -h /`
`df -hT`	Same, plus the filesystem type	`df -hT`
`df -i`	Inode usage (a disk can be "full" of tiny files)	`df -i /`
`du -sh DIR`	Total size of a directory	`du -sh ~/Downloads`
`du -h -d 1 DIR \| sort -h`	Size of each subdirectory, sorted	`du -h -d 1 ~ \| sort -h`
`du -sh * \| sort -h`	Size of everything here, sorted	`du -sh /var/log/* \| sort -h`
`ncdu DIR`	Interactive explorer; delete with `D`	`ncdu ~`
`lsblk`	Block devices and where they're mounted	`lsblk -f` (with filesystems and UUIDs)

Find big or old files:

find ~ -type f -size +500M -exec ls -lh {} +       # files over 500 MB
find /var/log -name '*.gz' -mtime +30              # compressed logs older than 30 days
find . -type f -printf '%s %p\n' | sort -rn | head # 10 largest files, in bytes

Links¶

A hard link is a second name for the same inode (the same data). A symbolic link (symlink) is a small file that points to a path.

Command	What it does	Example
`ln -s TARGET LINK`	Create a symlink	`ln -s /opt/app-2.1 /opt/app`
`ln -sfn TARGET LINK`	Repoint an existing symlink	`ln -sfn /opt/app-2.2 /opt/app`
`ln TARGET LINK`	Create a hard link (same filesystem, files only)	`ln data.csv data-link.csv`
`readlink LINK`	Show where a symlink points	`readlink /opt/app`
`readlink -f PATH`	Resolve all symlinks to a final absolute path	`readlink -f /usr/bin/vi`
`ls -li`	Show inode numbers (hard links share one)	`ls -li data*.csv`
`stat -c %h FILE`	Number of hard links	`stat -c %h data.csv`
`find -samefile F`	Find all hard links to a file	`find ~ -samefile data.csv`
`find -xtype l`	Find broken symlinks	`find ~ -xtype l`

-rw-rw-r-- 2 alex alex 6 Oct  2 10:45 data.csv        ← link count 2: two names, one inode
lrwxrwxrwx 1 alex alex 8 Oct  2 10:45 latest -> data.csv   ← symlink: type "l", shows its target

	Hard link	Symbolic link
Points to	An inode (the data)	A path (a name)
Works across filesystems	No	Yes
Can link to directories	No	Yes
If the original is deleted	Data survives	Link breaks ("dangling")

Common mistake: symlink argument order

It's ln -s TARGET LINK_NAME, the same order as cp SOURCE DEST. A relative target is resolved from the link's directory, not from where you ran the command.

Finding files fast¶

Command	What it does	Example
`find DIR -name PAT`	Search by name (quote the pattern)	`find ~ -name '*.csv'`
`find DIR -iname PAT`	Case-insensitive name	`find . -iname 'readme*'`
`find DIR -type d`	Directories only (`f` files, `l` symlinks)	`find /etc -type d -name '*.d'`
`find DIR -mtime -1`	Modified in the last 24 hours	`find ~/projects -mtime -1`
`find ... -exec CMD {} +`	Run a command on the results	`find . -name '*.tmp' -exec rm {} +`
`locate NAME`	Search a prebuilt index (fast, may be stale)	`locate sshd_config`
`which CMD`	Where a command lives in `PATH`	`which python3`
`type CMD`	What the shell will run: alias, builtin, or file	`type ls cd`