How to Use tar and gzip for Archiving and Compression

What Are tar and gzip?

#

tar (tape archive) bundles multiple files and directories into a single file while preserving permissions, ownership, and directory structure. It does not compress by itself.

gzip compresses a single file to reduce its size. It doesn’t handle multiple files or directories on its own.

Together they’re the standard approach on Linux: tar bundles the files, gzip compresses the bundle. The result is the familiar .tar.gz (or .tgz) file.

tar Basics

#

Create an archive

#

tar -cf archive.tar file1.txt file2.txt directory/

Extract an archive

#

tar -xf archive.tar

List contents without extracting

#

tar -tf archive.tar

Extract to a specific directory

#

tar -xf archive.tar -C /opt/myapp/

Verbose output

#

Add -v to any command to see each file being processed:

tar -cvf archive.tar ./project/
tar -xvf archive.tar

gzip Basics

#

Compress a file

#

gzip file.txt
# Produces file.txt.gz (original is removed)

Decompress a file

#

gunzip file.txt.gz
# Restores file.txt (compressed file is removed)

# Or equivalently
gzip -d file.txt.gz

Keep the original file

#

gzip -k file.txt
# Produces file.txt.gz AND keeps file.txt

Check compression ratio

#

gzip -l file.txt.gz

Combining tar and gzip

#

tar can invoke gzip directly with the -z flag, so you don’t need to run them separately.

Create a compressed archive

#

tar -czf archive.tar.gz ./project/

Extract a compressed archive

#

tar -xzf archive.tar.gz

List contents of a compressed archive

#

tar -tzf archive.tar.gz

Other Compression Options

#

tar supports multiple compression algorithms via different flags:

# bzip2
tar -cjf archive.tar.bz2 ./project/

# xz
tar -cJf archive.tar.xz ./project/

# zstandard
tar --zstd -cf archive.tar.zst ./project/

Practical Examples

#

Back up a directory

#

tar -czf backup-$(date +%F).tar.gz ./myapp/

Extract a single file from an archive

#

tar -xzf archive.tar.gz path/to/specific/file.txt

The path must match exactly as shown in tar -tzf.

Exclude files or directories

#

tar -czf project.tar.gz ./project/ --exclude='*.log' --exclude='node_modules'

Archive and preserve permissions (as root)

#

sudo tar -czpf full-backup.tar.gz /etc/

The -p flag preserves permissions. This is the default when running as root, but explicit is clearer.

Stream an archive over SSH

#

# Send a directory to a remote host without creating a local file
tar -cz ./project/ | ssh user@server 'tar -xz -C /opt/'

Create an archive with a top-level directory

#

# Puts everything under a project-v1.0/ prefix
tar -czf release.tar.gz --transform='s,^\./,project-v1.0/,' ./src/ ./README.md

Split a large archive into parts

#

tar -czf - ./large-dir/ | split -b 100M - backup-part-

# Reassemble and extract
cat backup-part-* | tar -xzf -

Common File Extensions

#

Verifying Archive Integrity

#

# Test that a gzip file is valid
gzip -t archive.tar.gz

# Verify tar contents are readable
tar -tzf archive.tar.gz > /dev/null

Troubleshooting

#

“tar: Removing leading ‘/’ from member names”

#

tar strips absolute paths by default to prevent accidentally overwriting system files on extraction. This is intentional and safe.

To extract to the original absolute paths (dangerous):

tar -xzf archive.tar.gz -C / --absolute-names

“gzip: stdin: not in gzip format”

#

The file isn’t actually gzip-compressed. Check what it is:

file archive.tar.gz

Permission errors during extraction

#

# Extract as yourself, ignoring stored ownership
tar -xzf archive.tar.gz --no-same-owner

Best Practices

#

• Use tar -czf as your default for creating archives — gzip is fast and universal
• Use xz or zstd when archive size matters more than speed
• Always list contents (tar -tzf) before extracting archives from untrusted sources
• Include a top-level directory in archives so extraction doesn’t scatter files into the current directory
• Use --exclude to leave out build artifacts, logs, and dependency directories
• Add a date stamp to backup filenames: backup-$(date +%F).tar.gz
• For large transfers, pipe tar over SSH instead of creating intermediate files