Networking setup and utilities on Linux


You probably already know these, in which case here's a reminder.

DNS Domain Name Service
PPP Point to Point Protocol
DHCP Dynamic Host Configuration Protocol
IP Internet Protocol
LAN Local Area Network
ISP Internet Service Provider
FTP File Transfer Protocol
SFTP Secure File Transfer Protocol


Linux became a low-cost and reliable server environment partly as a consequence of the availability of free software implementing all significant Internet protocols. An introduction to programming on Linux would leave a large gap without mention of popular networking utilities which assist with the process of setting up networks and network programming, and used for remote login and file transfer. Many engineers and students now want to setup Linux themselves, so information is provided to help you get your Linux installation networked.

Many network services are used in connection with Unix and Linux, for providing networked file services, logins shared across a network, providing mail and web services etc. Setting these up may involve installing and configure programs such as:

Linux is also often used for complex networking setups, e.g. for routers or firewalls involving multiple network cards and zones. These possibilities require building on a foundation of more basic knowledge which can partly be acquired by successfully connecting single machines to simple networks. Designing networks and diagnosing network faults are skilled actitities. These notes are intended to help you get started using Linux within this context.

Setting up networking on Linux

The 3 most likely scenarios are:
  1. Connecting to the Internet with a dial up modem using point-to point protocol (PPP).
  2. Using an Ethernet adaptor connected to a broadband modem.
  3. Connecting to the Internet using an Ethernet LAN, which has its own Internet connection, e.g. through a router.

Cases 2 and 3 generally require the same setup on the PC. In most cases, assuming that you have all the information needed to make a connection, the network configuration wizard software which comes with your Linux distribution will ask you to input this information at installation time, and this can be changed using helpful graphical programs such as Mandrake Control centre, Netconf etc.

ISPs, Linux and obtaining support

Some ISPs only officially support connection using closed source software on a CD which doesn't work on Linux. If you need to connect through such an ISP you might try asking in suitable Linux support mailing lists or newsgroups to find out if any of their Linux-using customers can give you any setup information or hints required, so you can find out the rest yourself. You are more likely to get useful help from newsgroup and list questions if you first make sure that your question is within the topic area of the group, and secondly try to find out as much as you reasonably can yourself e.g. if someone else has already asked the same question and this is searchable in the list archives or using the Google search engine. Those who might answer your questions are more likely to be able to give you useful answers if you are also willing to do some reading of the relevant Howto's, FAQs and manual pages which are readily accessible, and state what experiments you have tried yourself.

In almost all cases Internet connections are made using standard and open network protocols fully supported by Linux, (e.g. DHCP which allocates an Internet address from a pool every time you dial up). There are also many ISPs that explicitly support Linux users. In some cases you are better off as a customer of an ISP that can offer such support. Various links are available.

Information you are likely to need

If you are using a dial up modem you will need the remote phone number of the modems at your Internet service provider. They will also allocate you a username and password when you subscribe to their service. Many ISPs expect you to sign up to obtain your userid and password on-line using their web sites.

If you connect using DHCP, most of the information that might otherwise be needed is likely to be supplied and configured automatically when you connect. Creating a DHCP connection is often described on the networking setup wizards as "obtain network address automatically from a server" or similarly. Other information likely to be required includes:

  1. The addresses of DNS resolution servers. Usually this service will be backed up in which case 2 addresses need to be input. If you have to configure this information manually, it typically goes into /etc/resolv.conf .
  2. The address or name of the outgoing mail server provided by the ISP to their customers. It is likely to be much more efficient and reliable to send outgoing mail through this server even if you can send your outgoing mail directly from your own machine to the server handling mail for the destination address.
  3. The Internet (IP) address for your own PC. If you don't obtain the IP address from a server (using DHCP) you will need to input this address.
  4. The default gateway address. This is the address of the router which connects your LAN or the ISPs LAN to the Internet.
  5. The netmask configures the division between the network part of IP addresses and the host part. E.G. if your netmask is then the first 24 bits are the address of the destination network address and the last 8 are the host address. This is used to decide whether to send an outgoing network packet to the gateway or if it is intended for the local LAN. On some networks, e.g. using PPP (Point to Point Protocol) a netmask isn't needed.

If it doesn't work out of the box ...

If you use an up-to date quality Linux distribution, and have avoided buying the latest and cheapest network adaptors or "Win" modems which don't yet have a good level of Linux support you are less likely to have such difficulties, but occasionally you may need to dig into the source code and manual configuration files.

In most simple cases the network can successfully be installed by supplying the correct information as described above, either when Linux is installed, or by rerunning the network configuration wizard provided with your distribution (e.g. Netconf). In cases where things don't work, if you are operating within a LAN environment, it is worth checking that other parts of the network, (e.g. hubs, cables, routers and servers) are functional with other PCs if possible.

Sometimes you may need a non-standard setup, which can require you to carry out minor edits to scripts and/or configuration files. If you have to do this experimentally, always take dated backups of the files you are going to change first, so that you can restore these if the changes you try don't seem to have the desired effect.

If you are using a dial-up modem you are likely to need to slightly modify the default chat script used to dial your modem up. See chat(8) for details. Some of the configuration for PPP networking is in /etc/ppp . It is also worth studying the scripts in /etc/sysconfig which relate to networking and any associated manual pages you can find . On most Linux systems, system services are started by using shell scripts in /etc/rc.d and /etc/rc.d/init.d or similar pathnames . Most of these startup scripts can be run with parameters: start, stop and restart e.g:

cd /etc/rc.d/init.d
./network restart

If your Linux kernel was unable to identify drivers for network cards you are likely to get some useful clues by studying the boot message transcript obtainable using the dmesg(8) command.

Here is a section of the boot transcript obtained using dmesg showing details of the loading of a device driver for a 3COM PCI Ethernet card:

3c59x: Donald Becker and others.
See Documentation/networking/vortex.txt
02:02.0: 3Com PCI 3c905C Tornado at 0xa400. Vers LK1.1.18-ac
 00:01:01:f7:3d:21, IRQ 22
  product code 87cd rev 00.14 date 01-02-65
  Internal config register is 3800000, transceivers 0xa.
  8K byte-wide RAM 5:3 Rx:Tx split, autoselect/Autonegotiate interface.
  MII transceiver found at address 24, status 782d.
  Enabling bus-master transmits and whole-frame receives.
02:02.0: scatter/gather enabled. h/w checksums enabled

Using ping to diagnose network problems

ping(8) is typically used to evaluate whether a route exists to a host, and whether that host replies to ICMP (Internet Control Message Protocol) packets. E.G:

copsewood$ ping
PING ( 56(84) bytes of data.
64 bytes from icmp_seq=1 ttl=64 time=0.460 ms
64 bytes from icmp_seq=2 ttl=64 time=0.440 ms
64 bytes from icmp_seq=3 ttl=64 time=0.432 ms

--- ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 1998ms
rtt min/avg/max/mdev = 0.432/0.444/0.460/0.011 ms

To stop the sequence of packets being sent, <CTRL%gt; and <c%gt; keys were pressed simultaneously. I could have sent exactly 3 pings instead using the command: ping -c 3 .

You can use ping to also test whether your DNS setup works e.g:

copsewood$ ping
PING ( 56(84) bytes of data.
64 bytes from ( icmp_seq=1 ttl=113 time=109 ms
64 bytes from ( icmp_seq=2 ttl=113 time=106 ms
64 bytes from ( icmp_seq=3 ttl=113 time=105 ms

--- ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2023ms
rtt min/avg/max/mdev = 105.883/107.268/109.349/1.544 ms

Clearly if you can ping an IP address, but not the name of the same host, this would indicate that low-level routing was working as far as the host concerned, but that DNS was not.

Using ifconfig to check interface configurations

As root user, you can also use /sbin/ifconfig , to setup an experimental interface configuration see ifconfig(8). E.G:

ifconfig eth0

Sets the first ethernet adaptor to receive and send traffic using IP address: . You can use /sbin/ifconfig from any userid without parameters to check the status of the network and loopback interfaces:

copsewood$ /sbin/ifconfig
eth0      Link encap:Ethernet  HWaddr 00:01:01:F7:3D:21
          inet addr:  Bcast:  Mask:
          RX packets:1394045 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1393996 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100
          RX bytes:219979223 (209.7 Mb)  TX bytes:198606294 (189.4 Mb)
          Interrupt:22 Base address:0xa400

lo        Link encap:Local Loopback
          inet addr:  Mask:
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:347 errors:0 dropped:0 overruns:0 frame:0
          TX packets:347 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:34059 (33.2 Kb)  TX bytes:34059 (33.2 Kb)

If an inet addr has not been configured this would indicate a problem with network startup when the PC booted.

Using telnet

telnet(1) was a service allowing for remote shell login from a user on one computer (the client) to access services provided by another (the server). This provided a platform for remote access to a variety of interactive text-based public services, e.g. library catalogues prior to the existence of the world-wide web. Since then, to the extent that public access is needed to well-defined system functions or information, HTTP has provided a more usable protocol. However, where access is required privately by programmers or systems administrators to facilities on a remote computer less susceptible to time- consuming design and definition as web pages and web-accessible data and services, the need for remote login and remote control facilities has continued.

Traditional telnet is no longer believed to be secure enough to carry out this task over the Internet, given the greater probability of surveillance by those interested in exploiting secrets which can be sniffed from a plain-text network interaction. Consequently, for the purposes for which a remote shell login application is required, Secure Shell ssh(1) has effectively replaced telnet.

Telnet is, however, still useful for debugging text mode network protocols, where you are trying to find out what a server is doing. In the following dialogue, a SMTP mail relay was accessed on TCP port 25 to see whether it's relay backup function was correctly setup for a particular domain.

  [rich@february rich]$ telnet 25
Connected to (
Escape character is '^]'.
MAIL From:
250 ok
250 ok
354 go ahead
Subject: testing the new MX relay

a message
250 ok 1068112145 qp 7650
Connection closed by foreign host.

ssh secure shell

ssh uses a technique known as public key encryption to authenticate the originator of the connection to the server and also to authenticate the server to the client. After key exchange, the connection is carried out over an encrypted tunnel.

ssh can be setup to require a password, or in cases where the connection has to be automated, simple automated exchange of keys. A trust relationship has to be established first. This is achieve when the server knows the validity of the client public key and the client knows the validity of the server key. When you first make a connection from a ssh client to ssh server, you will be asked to confirm the identity of the server key, to make it more difficult for a bogus server to steal your password or other details using a "man in the middle" attack.

To use ssh on a server to allow remote login etc. you will need to install sshd, or request that this package be installed at installation time. Use the ps -fe command to check whether sshd is running.

From the client use the command:

ssh userid@server_address

to initiate a secure shell login. You will need to specify your real userid on the server, and either the DNS domain or the IP address for the server_address. If this is the first time you have connected to the specific server you will be asked to confirm its public key fingerprint. Details of these public keys are stored in ~/.ssh/known_hosts . To finish a shell login use the exit or logout commands.

Starting GUI software from a ssh session over X-Windows

This is not recommended over a slow network connection, due to the bandwidth requirements of GUI software. However, it is often useful to run an application on a server which is displayed on a client. If the application and client use X-Window protocols (all modern Unixes, Linux etc) this is possible.

The display of an application running on the server is redirected by changing the DISPLAY environment variable to the client display.

-bash-2.05b$ echo $DISPLAY
-bash-2.05b$ DISPLAY=
-bash-2.05b$ export DISPLAY
-bash-2.05b$ which konqueror
-bash-2.05b$ konqueror &

This session caused the KDE file manager (konqueror) to run on the ssh server with a GUI window displayed on and controllable by the ssh client PC.

Using ftp to transfer files

ftp(1) is used for anonymous and private file transfer, and for uploading files to web servers. For all but anonymous public file access ftp has become insecure for the same reasons that telnet is now obsolete. However, many low-cost or free web-hosting companies reason that ftp reduces their support costs as it is so widely used, and it is only their users data and websites which are at risk. If you need to transfer files onto a system on which you have a ssh session e.g. to upgrade a software package on a remote server, command line ftp is likely to be useful. Anonymous ftp uses the ftp username: anonymous and the password, by convention is your email address.

Transfer of files using ftp allows for the regular and automated synchronisation of large data collections.

Anonymous ftp is very widely used to distribute freely available software packages. Here are selected parts of a ftp session transcript:

[rich@copsewood rich]$ ftp
Connected to
220 UK Mirror Service FTP server version 1.4 ready
Name ( anonymous
331 Guest login ok: please send your email address as the password
230-  Welcome to the UK Mirror Service funded by JISC
230-  More information can be found at our web site:
230-  Please send comments or questions to
230 Logged in for anonymous FTP
Remote system type is UNIX.
Using binary mode to transfer files.
ftp> ls
227 Entering Passive Mode (212,219,56,152,237,212)
150 About to open data connection
drwxr-xr-x    1 ukms     ukms         1024 Oct 29 00:00 collections
drwxrwxr-x    1 ukms     ukms         1024 Oct 29 08:08 indexes
drwxr-xr-x    1 ukms     ukms          512 Oct 29 07:03 pub
226 Transfer complete
ftp> cd indexes
250 Current directory now /indexes
ftp> ls
227 Entering Passive Mode (212,219,56,152,238,9)
150 About to open data connection
drwxrwxr-x    1 ukms     ukms         8704 Oct 26 08:05 by-mirror
-rw-r--r--    1 ukms     ukms     29040129 Oct 29 08:08 ls-lR.gz
-rw-r--r--    1 ukms     ukms            0 Oct  7 08:05
226 Transfer complete
ftp> cd by-mirror
250 Current directory now /indexes/by-mirror
ftp> ls
-rw-r--r--    1 ukms     ukms         5245 Oct 26 01:01
ftp> binary
200 Type set to I
226 Transfer complete
ftp> get
local: remote:
150 About to open data connection for file (5245 bytes).
226 Transfer complete
5245 bytes received in 0.056 seconds (91 Kbytes/s)
ftp> quit
221 Goodbye
[rich@copsewood rich]$

ftp has various subcommands. cd and ls have the same purpose as within a local shell, but on the remote system. put and get are used to copy a file from the client to the server or from the server to the client. These work in a similar manner to the cp shell command, but you don't need to specify the target name if you are happy to have the same filename on the transferred file. help obtains help information and quit and/or exit will terminate your ftp session. You may need to use binary and ascii commands to switch between binary and text transfer modes. Text transfer involves converting line ending characters in the transferred file if the operating system text conventions are different on the client and server. The hash ftp command was sometimes useful with large files over slow connections as it allows the speed of transfer to be monitored.

Transferring files securely using sftp(1)

For secure file transfer, using private logins and passwords the sshd server will also transfer files between client and server using the sftp(1) command, which works in much the same way as ftp. One difference is that you will need to start the session using a command of the form:

sftp userid@server

instead of ftp server and then being prompted for userid.

Another useful difference is that you don't need to run a seperate server for sftp service. The secure shell daemon: sshd also provides sftp file transfers.

Other useful network utility programs

traceroute(8) is useful for investigating routing problems to a host. dig(1) is used to check DNS records for the Domain Name Service.

Linux is typically installed with a range of conventional GUI-based programs for network browsing and email such as Mozilla and Evolution. For some purposes, e.g. when accessing a remote system using ssh, the faster and smaller text-based tools can be used. These are also worth investigation if you want to automate operations involving mail or web page access, e.g. within shell scripts.

lynx(1) text mode web browser
mutt(1) powerful email client for fast mailbox acess etc.
mail(1) an ancient mail program useful in shell scripts and commands.