netcat-tutorial

Netcat is a simple but handy UNIX utility that reads and writes data across network connections, using either TCP or UDP. The purpose of this tutorial is to help you learn the netcat command line utility and use it productively.

Before You Begin

Some of the commands in this guide will require the use of two terminal windows running netcat, one acting as a server and the other as the client. These can be separate machines, or you can connect to the same localhost.

If you are running a Linux Distribution, consider using Terminator or tmux

This guide is written for a non-root user. Depending on your configuration, some commands might require the help of sudo in order to get property executed.

Port numbers 0-1024 are restricted and can only be used with root privileges, which means that you should use the sudo command for creating TCP/IP servers that use port numbers 0-1024. This rule does not apply to TCP/IP clients that use port numbers 0-1024.

Introduction

As netcat is not installed by default, you will most likely need to install netcat on your Linux machine using your favorite package manager.

The netcat binary usually has an alias named nc, which is what will be used in this guide because it is shorter. Usually both commands point to the same binary file.

If you execute apt-cache search netcat on a Debian machine, you will see the following output:

apt-cache search netcat

netcat - TCP/IP swiss army knife -- transitional package
netcat-traditional - TCP/IP swiss army knife
netcat-openbsd - TCP/IP swiss army knife

Notice that netcat is a dummy package and its purpose is to ease upgrades. The differences between the other two packages are not big, but you will need to visit their package descriptions and their man pages in order to get a detailed description of their capabilities.

The OpenBSD version of netcat supports IPv6, proxies and UNIX sockets, which are not supported by the netcat-traditional variant. On the other hand, netcat-traditional includes support for the -e option that allows you to execute a program from a remote shell, which is not offered by netcat-openbsd. However, if you do not need any of these features, you will not notice any real difference between these two versions of netcat.

Command Line Options

netcat commands have the netcat [options] host port generic form. The nc binary supports the following command line options:

The remainder of this guide will demonstrate the most important of these commands. That being said, netcat is a versatile tool, and there’s a large opportunity for experimenting on your own.

Using netcat as a Client

The most common use of netcat is to act as a client for a server process. This is mostly used for troubleshooting network servers and network connections because you can see the raw data of the interaction. So, providing nc with just a hostname or IP address and a port number will make netcat act as the telnet utility:

nc localhost 22

SSH-2.0-OpenSSH_7.9p1 Debian-10

In the given example, nc tries to connect to TCP port number 22 of the localhost - notice that TCP port number 22 is used by SSH, which is what triggers the provided output.

Also notice that as the -u option is not used, nc will use the TCP protocol by default.

Using netcat as a Server

nc will accept connections at a given port and act as a server when you execute it with the -l option:

nc -l -p 1234

In another terminal window, connect a client to the server with nc:

nc 127.0.01 1234

You can now send messages between the two machines with nc.

This is a client!
Hello from the server!

The previous command tells netcat to listen on TCP port number 1234 for incoming connections - you can also see that netcat automatically reads data from the client and that you can send your response to the TCP client just by typing it.

Once again, as the -u option is not used, nc will use the TCP protocol.

Getting Verbose Output

There are times where you cannot connect to the remote machine or the answer you get is not the expected one. In such cases, it is good to use either -v or -vv in order to get more information from the nc connection.

nc -v localhost 1234

localhost [127.0.0.1] 1234 (?) : Connection refused

The output you get shows that the reason you cannot connect to localhost using TCP port number 1234 is that your connection was refused by the server. Executing nc localhost 1234 will return no output, which offers no help.

Using -vv instead of -v will generate the following kind of output:

nc -vv localhost 1234

localhost [127.0.0.1] 1234 (?) : Connection refused
 sent 0, rcvd 0

If the TCP connection was successful, you would have gotten the following kind of output on the client side:

nc -vv localhost 1234

localhost [127.0.0.1] 1234 (?) open

Both -v and -vv are very valuable when things do not work as expected.

Using the UDP Protocol

In order to use the UDP protocol instead of the TCP protocol, you should include the -u option in your nc commands. Therefore, the following command will use the UDP protocol:

nc –vv –u 8.8.8.8 53

dns.google [8.8.8.8] 53 (domain) open

As we are trying to connect to a (public) DNS server, we will have to use port number 53.

Examples

In this section you will find a number of use cases and examples for nc.

Using netcat for Port Scanning

Netcat can be used for port scanning as a naïve version of nmap with the -z option. The command that follows scans the localhost, which has an IP address of 127.0.0.1, using a range of port numbers from 1 to 30 (1-30):

netcat -z -vv -n 127.0.0.1 1-30

(UNKNOWN) [127.0.0.1] 30 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 29 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 28 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 27 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 26 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 25 (smtp) open
(UNKNOWN) [127.0.0.1] 24 (?) : Connection refused
(UNKNOWN) [127.0.0.1] 23 (telnet) : Connection refused
(UNKNOWN) [127.0.0.1] 22 (ssh) open
(UNKNOWN) [127.0.0.1] 21 (ftp) : Connection refused
(UNKNOWN) [127.0.0.1] 20 (ftp-data) : Connection refused
(UNKNOWN) [127.0.0.1] 19 (chargen) : Connection refused

Notice that as we are using the -n option, the server should be specified by its IP address. Additionally, if you omit the -vv option, you will get a much shorter output, which is verified by the following output:

nc -z -v -n 127.0.0.1 1-30

(UNKNOWN) [127.0.0.1] 25 (smtp) open
(UNKNOWN) [127.0.0.1] 22 (ssh) open

Therefore, the use of -v makes nc to display open TCP ports only.

If you do not use -v or -vv, the previous command will return no output:

nc -z -n 127.0.0.1 1-30

Using netcat for Transferring Files

One of the features of netcat is that it is capable of transferring files:

cat access.log | nc -vv -l -p 4567

listening on [any] 4567 ...
connect to [127.0.0.1] from localhost [127.0.0.1] 53952

When a client connects to TCP port number 4567, nc will send the contents of the access.log file to it. The correct way to execute a nc client in order to get that file is the following. Open a new terminal window and enter this command:

nc -vv localhost 4567 > fileToGet

localhost [127.0.0.1] 4567 (?) open
sent 0, rcvd 362148
^C

You will need to press Control+C for the TCP connection to close.

Using netcat for Making any Process a Server

Netcat allows you to make any process a server process with the help of the –e parameter:

nc -vv -l -p 12345 -e /bin/bash

listening on [any] 12345 ...
connect to [127.0.0.1] from localhost [127.0.0.1] 46930
bash: line 2: asd: command not found

Here you tell nc to accept incoming TCP connections on TCP port number 12345. When a connection is accepted, nc will execute /bin/bash, which means that it will give you shell access on the machine. After a client successfully connects, every input line will be executed as a shell command using /bin/bash. If the command cannot be found, the client will get no output and an error message will be generated on the server side. Otherwise, the output of the command will be sent to the client. To test this functionality, in another terminal window create a nc client and type in the following command:

nc localhost 12345

This capability of netcat can introduce security threats on your Linux machine when used improperly. It is advised that you exercise caution if using this feature.

Executing a Command After Connecting

If you want to execute a given command each time a client connects to a server that is implemented using nc, then you should use the -c option followed by that command. The example that follows executes ls -l and sends the output to the client:

nc -vv -c "ls -l" -l 127.0.0.1 -p 1234

listening on [any] 1234 ...
connect to [127.0.0.1] from localhost [127.0.0.1] 33788

Try executing nc 127.0.0.1 1234 on another terminal on your local machine to get the output of ls -l.

This capability of netcat can introduce security threats on your Linux machine when used improperly. It is advised that you exercise caution if using this feature.`

Using netcat as a Simple Web Server

Let us say that you want to serve a simple HTML page, which in this case will be called index.html, from your Linux machine but you have no real web server available. You can use netcat to serve that simple HTML page on clients from your local machine as follows:

nc -vv -l 127.0.0.1 -p 4567 < index.html

Using wget to get that page will generate the following output in the nc part:

listening on [any] 4567 ...
connect to [127.0.0.1] from localhost [127.0.0.1] 53980
GET / HTTP/1.1
User-Agent: Wget/1.18 (linux-gnu)
Accept: */*
Accept-Encoding: identity
Host: localhost:4567
Connection: Keep-Alive

Additionally, when using wget, we will receive the following output, which reflects the contents of the index.html page:

wget -qO- http://localhost:4567/

<html>
    <head>
    <title>Page Under Construction</title>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=UTF-8">
<style type="text/css">
body {
  background-color:#000;
}
.img {
  margin:180px 50px 75px 450px;
}
    </style>
   </head>
<body&>
    <H1>
        Demo Page
    </H1>
    </body>
</html>

Using netcat for Getting Data from Web Servers

The HTTP service is just a TCP service; therefore nc can be used for getting data from a web server or for testing web servers. The following command will connect to the www.example.com machine using port number 80, which corresponds to the HTTP protocol:

nc www.example.com 80

You should type the first line (GET / HTTP/1.1) and press the enter key two times in order to get a response from the web server.

GET / HTTP/1.1

HTTP/1.1 400 Bad Request
Server: nginx
Date: Wed, 10 Jul 2019 20:02:47 GMT
Content-Type: text/html
Content-Length: 166
Connection: close

<html>
<head>
<title>400 Bad Request</title>
</head>
<body bgcolor="white">
<center><h1>400 Bad Request</h1>/center>
</body>
</html>

A better way to execute this command is the following:

echo -en "GET / HTTP/1.0\n\n\n" | netcat www.example.com 80

HTTP/1.1 301 Moved Permanently
Server: nginx
Date: Wed, 10 Jul 2019 20:04:10 GMT
Content-Type: text/html
Content-Length: 178
Connection: close
Location: https:///
<html>
<head><title>301 Moved Permanently</title></head>
<body bgcolor="white">
<center><h1>301 Moved Permanentl</h1<;/center>
<hr><center>nginx</center>
</body>
</html>

This used to be a very popular way of testing web servers when every web server was using the HTTP protocol. Nowadays, the use of HTTPS makes difficult to test a web server using tools such as netcat and telnet because the web traffic is encrypted.

Using netcat for Creating a Chat Server

Creating a basic chat server with nc for two machines to communicate with each other is completed in two commands. One of the machines will function as the server and the other machine will be the client. On the server you will need to execute the following:

nc -vv -l 127.0.0.1 -p 1234

listening on [any] 1234 ...
connect to [127.0.0.1] from localhost [127.0.0.1] 60608
Hello!

And on the client:

nc -vv 127.0.0.1 1234

Hello!

If both people that want to talk are on the same Linux machine, then using 127.0.0.1 is safer and quicker. Otherwise, you should use the IP address of the server in both commands.

Transferring Entire Directories Using netcat

This section will explain how to transfer entire directories using netcat. Imagine that you wish to transfer the var directory that resides under your home directory. You can do that as follows:

tar -cvf - ~/var | nc -vv -l 127.0.0.1 -p 1234

listening on [any] 1234 ...
tar: Removing leading `/' from member names
/home/username/var/
/home/username/var/slide.tar.ORG

This creates a TCP server that listens on TCP port number 1234 on the host with the 127.0.0.1 IP address. Generally speaking, using 127.0.0.1 as the server IP address is more secure than using one of the real IP addresses of your Linux machine provided that both the server and the client are on the same Linux machine.

After that you will need to execute the following command on the client side:

cd /tmp
nc 127.0.0.1 1234 | tar -xvf -

home/username/var/
home/username/var/slide.tar.ORG
home/username/var/after.tshark
home/username/var/test.pcap
home/username/var/sys09725827.php
home/username/var/test.php
home/username/var/u5EJqp.php
home/username/var/http.pcap
home/username/var/sketch.zip

When the client connects, the nc server will also print the following output:

listening on [any] 1234 ...
tar: Removing leading `/' from member names
/home/username/var/
/home/username/var/slide.tar.ORG
connect to [127.0.0.1] from localhost [127.0.0.1] 60632
/home/username/var/after.tshark
/home/username/var/test.pcap
/home/username/var/sys09725827.php
/home/username/var/test.php
/home/username/var/u5EJqp.php
/home/username/var/http.pcap
/home/username/var/sketch.zip
 sent 3645440, rcvd 0
 ^C

You will need to press Control+C for the TCP connection to close.

Testing the Network Speed Using netcat

This section will explain how to test the connection speed between two machines using nc. You will need two hosts. On the server machine use the following command:

time nc -vv -n -l -p 2222 >/dev/null

listening on [any] 2222 ...
connect to [127.0.0.1] from (UNKNOWN) [127.0.0.1] 42286
 sent 0, rcvd 2090934272

real    0m21.438s
user    0m0.230s
sys 0m1.190s

On the client machine, you should execute the following command and press Control+C after the desired amount of time to end the connection:

time yes | nc.traditional -vv -n 127.0.0.1 2222 >/dev/null

(UNKNOWN) [127.0.0.1] 2222 (?) open
^C sent 2090926080, rcvd 0

real    0m5.482s
user    0m0.456s
sys 0m3.109s

Now that you know it took 5.482s to transfer 2090926080 bytes, you can calculate the network speed. As the nc server starts first, you should use the numbers found in the nc client.