wget — quick reference
Basic downloads
Fetch a URL to the current directory or a path you choose. Examples use https://example.com — a safe documentation domain.
| When to use | Command |
|---|---|
| Download using the remote filename | wget https://example.com |
| Save under a specific filename | wget -O page.html https://example.com |
| Save into a directory (creates it if needed) | wget -P /tmp/dl https://example.com |
| Check that a URL exists without saving body | wget --spider -q https://example.com |
| Print server response headers | wget -S --spider https://example.com |
Resume, retries, and rate limits
| When to use | Command |
|---|---|
| Continue a partial download | wget -c https://example.com/large.iso |
| Retry up to 10 times on failure | wget -t 10 https://example.com |
| Unlimited retries (0 means infinite) | wget -t 0 https://example.com |
| Cap download speed (example 500 KB/s) | wget --limit-rate=500k https://example.com |
| Wait 5 seconds between retries | wget --waitretry=5 -t 10 https://example.com |
Batch and background
| When to use | Command |
|---|---|
| Download every URL listed in a file | wget -i urls.txt |
| Quiet mode for scripts | wget -q -O out.html https://example.com |
| Log messages to a file | wget -o wget.log https://example.com |
| Run in the background | wget -b https://example.com |
Timestamping and clobber control
| When to use | Command |
|---|---|
| Skip download if local file is newer or same age | wget -N https://example.com/file |
| Do not overwrite an existing file | wget -nc -O page.html https://example.com |
Recursive and mirror (use carefully)
| When to use | Command |
|---|---|
| Recursive download (respect site rules) | wget -r -l 2 https://example.com/ |
| Mirror mode (timestamping + infinite depth by default) | wget -m https://example.com/ |
| Stay under one directory (no parent paths) | wget -r -np https://example.com/dir/ |
| Flatten paths into the current directory | wget -r -nd https://example.com/dir/ |
Authentication and TLS
| When to use | Command |
|---|---|
HTTP basic auth (prefer --ask-password) |
wget --user=USER --ask-password https://example.com/ |
| Custom header (API token) | wget --header="Authorization: Bearer TOKEN" -O out.json https://api.example.com/ |
| Skip TLS verification (lab only — insecure) | wget --no-check-certificate https://selfsigned.local/ |
wget — command syntax
Synopsis from wget --help on Ubuntu 25.04 (GNU Wget 1.24.5):
wget [OPTION]... [URL]...With no URL, wget can read links from -i or standard input. Options before URLs apply to all files retrieved in that run.
wget — command examples
Essential Download and custom filename
The simplest fetch stores index.html in the current directory. -O picks the local name explicitly — useful in scripts.
cd /tmp && rm -f example-dl.html
wget -q -O /tmp/example-dl.html https://example.com
ls -la /tmp/example-dl.html
head -c 60 /tmp/example-dl.html; echoSample output:
-rw-r--r-- 1 root root 559 Jul 1 02:26 /tmp/example-dl.html
<!doctype html><html lang="en"><head><title>Example Domain</title>Remove the test file when finished: rm -f /tmp/example-dl.html.
Essential Target directory and spider check
-P creates the destination directory. --spider checks reachability without writing the response body — handy in health checks.
rm -rf /tmp/wget-dl && mkdir -p /tmp/wget-dl
wget -q -P /tmp/wget-dl https://example.com
ls /tmp/wget-dl/
wget --spider -q https://example.com
echo "spider exit code: $?"
rm -rf /tmp/wget-dlSample output:
index.html
spider exit code: 0Exit code 0 means the URL responded successfully.
Common Server headers and retry count
-S with --spider prints response headers to stderr. -t limits retry attempts when the network flaps.
wget -S --spider https://example.com 2>&1 | head -10
wget -q -t 2 -O /tmp/wget-retry.html https://example.com && echo OK
rm -f /tmp/wget-retry.htmlSample output:
Spider mode enabled. Check if remote file exists.
HTTP request sent, awaiting response...
HTTP/1.1 200 OK
Content-Type: text/html
OKExact header lines depend on the remote server and any proxy.
Common Batch download from a URL list
Put one URL per line in a text file and pass -i. wget fetches each link in order.
rm -rf /tmp/wget-batch && mkdir -p /tmp/wget-batch
printf '%s\n' 'https://example.com' > /tmp/wget-batch/urls.txt
wget -q -i /tmp/wget-batch/urls.txt -P /tmp/wget-batch/out
ls /tmp/wget-batch/out/
rm -rf /tmp/wget-batchSample output:
index.htmlAdd -B / --base when the list contains relative links resolved against one site.
Common Quiet mode and log file
-q silences progress — pair with -O in cron jobs. -o keeps a log when you still need diagnostics.
wget -q -O /tmp/wget-quiet.html https://example.com
wc -c /tmp/wget-quiet.html
wget -o /tmp/wget.log -O /tmp/wget-logged.html https://example.com
tail -1 /tmp/wget.log
rm -f /tmp/wget-quiet.html /tmp/wget-logged.html /tmp/wget.logSample output:
559 /tmp/wget-quiet.html
2026-07-01 … FINISHED …The log line format includes timestamp and FINISHED on success.
Advanced Limit download rate
--limit-rate protects shared links or production bandwidth when pulling large artifacts.
wget -q --limit-rate=100k -O /tmp/wget-rate.html https://example.com
ls -la /tmp/wget-rate.html
rm -f /tmp/wget-rate.htmlSample output:
-rw-r--r-- 1 root root 559 … /tmp/wget-rate.htmlSuffix k or m sets kilobytes or megabytes per second as documented in wget --help.
Advanced Timestamping and no-clobber
-N skips re-downloading when the local copy is up to date. -nc refuses to overwrite an existing target.
wget -q -O /tmp/wget-ts.html https://example.com
wget -nc -q -O /tmp/wget-ts.html https://example.com 2>&1
echo "nc exit: $?"
rm -f /tmp/wget-ts.htmlSample output:
File '/tmp/wget-ts.html' already there; not retrieving.
nc exit: 0Without -nc, wget overwrites by default.
Advanced Shallow recursive fetch — use responsibly
Recursive mode can hammer remote sites. Limit depth with -l and scope with -np. Always respect robots.txt and site policy.
rm -rf /tmp/wget-mirror && mkdir -p /tmp/wget-mirror
wget -q -r -l 1 -np -P /tmp/wget-mirror https://example.com/
find /tmp/wget-mirror -type f | head -5
rm -rf /tmp/wget-mirrorSample output (paths vary):
/tmp/wget-mirror/example.com/index.htmlFor full offline mirrors, -m adds timestamping and recursion defaults — test on mirrors you own first.
wget — when to use / when not
| Use wget when | Use something else when |
|---|---|
| You download files or mirror static sites from scripts | You call REST APIs with JSON POST bodies → curl |
You need resume (-c) and simple retry knobs |
You upload multipart forms or need HTTP/2 client features → curl |
You batch hundreds of URLs with -i |
You already use a language HTTP library in application code |
You want unattended background downloads (-b) |
You need interactive browsing → elinks or a browser |
| You mirror directory trees with built-in recursion | You need torrent or metalink clients → specialized tools |
wget vs curl
| wget | curl | |
|---|---|---|
| Primary job | Non-interactive download and mirror | Data transfer and API testing |
| Recursive mirroring | Built-in (-r, -m) |
No native recursive crawl |
| Resume downloads | -c |
--continue |
| Output default | Saves to file | Prints body to stdout |
| Protocol breadth | HTTP, HTTPS, FTP focus | Many protocols and verbs |
| Best for | Cron downloads, ISO pulls, mirrors | Headers, POST, OAuth APIs |
See the curl command for API-oriented examples.
Related commands
| Command | One line |
|---|---|
| scp | Copy files over SSH |
| rsync | Efficient sync and transfers |
| tar | Extract archives after download |
Browse the full index in our Linux commands reference.
wget — interview corner
What is wget used for in Linux?
GNU wget retrieves files over HTTP, HTTPS, and FTP from the shell. It is non-interactive — ideal for scripts, servers without a GUI, and cron jobs. It follows redirects, retries failures, and can resume partial files with -c.
A strong answer is:
"wget is a non-interactive downloader for HTTP/HTTPS/FTP — I use it in scripts for ISOs and artifacts, with -c to resume and -q for quiet cron runs."
How do you resume an interrupted wget download?
Use -c / --continue. wget asks the server for the rest of the file when the protocol supports range requests.
wget -c https://example.com/large.isoIf resume fails, the server may not support ranges or the local file is corrupt — remove the partial file and restart.
A strong answer is:
"wget -c continues partial downloads when the server supports byte ranges; otherwise I delete the bad partial and restart."
When do you pick wget over curl?
Pick wget for file downloads, batch URL lists (-i), and mirroring static sites. Pick curl when you inspect headers, send POST/PUT with JSON, or debug APIs.
Both can download a single URL; wget defaults to saving a file, curl defaults to stdout.
A strong answer is:
"wget for bulk downloads and mirrors; curl for APIs and custom HTTP methods. wget -O file URL versus curl -o file URL for a single save."
How do you mirror a website with wget?
Mirror mode is wget -m URL (shortcut for recursion, timestamping, and more). In practice, add depth and scope limits:
wget -m -l 2 -np https://example.com/docs/Respect bandwidth, robots.txt, and legal terms — uncontrolled mirrors can overload small sites.
A strong answer is:
"wget -m for mirror semantics, but I cap depth with -l and scope with -np, and only mirror sites I am allowed to copy."
What does wget --spider do?
--spider performs the request without saving the response body — useful as a URL health check in monitoring scripts. Combine with -S to see response headers.
Exit status is 0 on success, non-zero on failure — testable in shell if statements.
A strong answer is:
"--spider checks reachability without writing a file; I use it in scripts and add -S when I need headers."
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
ERROR: certificate verification failed |
TLS trust or hostname mismatch | Fix CA bundle; lab-only: --no-check-certificate (insecure) |
| Resume re-downloads from zero | Server lacks range support | Remove partial file; try a mirror with ranges |
403 Forbidden |
Auth, hotlink block, or User-Agent filter | Add --user / headers; check site policy |
Connection refused |
Service down or wrong port | Verify URL; increase -t retries |
| Proxy errors | http_proxy set in environment |
Unset proxy or set --no-proxy for direct fetch |
| Recursive wget pulls entire internet | Missing -l / -np |
Add depth and domain limits; never run unbounded -r on production WAN |
Empty file with -O |
URL returned error page | Check exit code; use -S to read HTTP status |
File already there; not retrieving |
-nc and existing file |
Remove file or drop -nc |
