Basic Linux Networking Commands You Should Know


Brief: A collection of most important and yet basic Linux networking commands an aspiring Linux SysAdmin and Linux enthusiasts must know.

It’s not every day at It’s FOSS that we talk about the “command line side” of Linux. Basically, I focus more on the desktop side of Linux. But as some of you readers pointed out in the internal survey (exclusive for It’s FOSS newsletter subscribers), that you would like to learn some command line tricks as well. Cheat sheets were also liked and encouraged by most readers.

For this purpose, I have compiled a list of the basic networking commands in Linux. It’s not a tutorial that teaches you how to use these commands, rather, it’s a collection of commands and their short explanation. So if you already have some experience with these commands, you can use it for quickly remembering the commands.

You can bookmark this page for quick reference or even download all the commands in PDF for offline access.

I had this list of Linux networking commands when I was a student of Communication System Engineering. It helped me to get the top score in Computer Networks course. I hope it helps you in the same way.

Exclusive bonus: Download Linux Networking Commands Cheat Sheet for future reference. You can print it or save it for offline viewing.

List of basic networking commands in Linux

I used FreeBSD in the computer networking course but the UNIX commands should work the same in Linux also.


ping <host> —- sends an ICMP echo message (one packet) to a host. This may go continually until you hit Control-C. Ping means a packet was sent from your machine via ICMP, and echoed at the IP level. ping tells you if the other Host is Up.

telnet host <port> —- talk to “hosts” at the given port number. By default, the telnet port is port 23. Few other famous ports are:
7 – echo port,
25 – SMTP, use to send mail
79 – Finger, provides information on other users of the network

Use control-] to get out of telnet.


Arp is used to translate IP addresses into Ethernet addresses. Root can add and delete arp entries. Deleting them can be useful if an arp entry is malformed or just wrong. Arp entries explicitly added by root are permanent — they can also be by proxy. The arp table is stored in the kernel and manipulated dynamically. Arp entries are cached and will time out and are deleted normally in 20 minutes.

arp –a : Prints the arp table
arp –s <ip_address> <mac_address> [pub] to add an entry in the table
arp –a –d to delete all the entries in the ARP table


netstat –r —- Print routing tables. The routing tables are stored in the kernel and used by ip to route packets to non-local networks.
route add —- The route command is used for setting a static (non-dynamic by hand route) route path in the route tables. All the traffic from this PC to that IP/SubNet will go through the given Gateway IP. It can also be used for setting a default route; i.e., send all packets to a particular gateway, by using in the pace of IP/SubNet.
routed —– The BSD daemon that does dynamic routing. Started at boot. This runs the RIP routing protocol. ROOT ONLY. You won’t be able to run this without root access.
gated —– Gated is an alternative routing daemon to RIP. It uses the OSPF, EGP, and RIP protocols in one place. ROOT ONLY.
traceroute —- Useful for tracing the route of IP packets. The packet causes messages to be sent back from all gateways in between the source and destination by increasing the number of hopes by 1 each time.
netstat –rnf inet : it displays the routing tables of IPv4
sysctl net.inet.ip.forwarding=1 : to enable packets forwarding (to turn a host into a router)
route add|delete [-net|-host] <destination> <gateway> (ex. route add to add a route
route flush : it removes all the routes
route add -net : to add a default route
routed -Pripv2 –Pno_rdisc –d [-s|-q] to execute routed daemon with RIPv2 protocol, without ICMP auto-discovery, in foreground, in supply or in quiet mode
route add : it defines the route used from RIPv2
rtquery –n : to query the RIP daemon on a specific host (manually update the routing table)


nslookup —- Makes queries to the DNS server to translate IP to a name, or vice versa. eg. nslookup will gives you the IP of
ftp <host>water —– Transfer files to host. Often can use login=“anonymous” , p/w=“guest”
rlogin -l —– Logs into the host with a virtual terminal like telnet

Important Files:

/etc/hosts —- names to ip addresses
/etc/networks —- network names to ip addresses
/etc/protocols —– protocol names to protocol numbers
/etc/services —- tcp/udp service names to port numbers

Tools and network performance analysis

ifconfig <interface> <address> [up] : start the interface
ifconfig <interface> [down|delete] : stop the interface
ethereal & : it allows you open ethereal background not foreground
tcpdump –i -vvv : tool to capture and analyze packets
netstat –w [seconds] –I [interface] : display network settings and statistics
udpmt –p [port] –s [bytes] target_host : it creates UDP traffic
udptarget –p [port] : it’s able to receive UDP traffic
tcpmt –p [port] –s [bytes] target_host : it creates TCP traffic
tcptarget –p [port] it’s able to receive TCP traffic

netmask [up]

: it allows to subnet the sub-networks


ifconfig sl0 srcIP dstIP : configure a serial interface (do “slattach –l /dev/ttyd0” before, and “sysctl net.inet.ip.forwarding=1“ after)
telnet : to access the switch from a host in its subnetwork
sh ru or show running-configuration : to see the current configurations
configure terminal : to enter in configuration mode
exit : in order to go to the lower configuration mode


vlan n : it creates a VLAN with ID n
no vlan N : it deletes the VLAN with ID N
untagged Y : it adds the port Y to the VLAN N
ifconfig vlan0 create : it creates vlan0 interface
ifconfig vlan0 vlan ID vlandev em0 : it associates vlan0 interface on top of em0, and set the tags to ID
ifconfig vlan0


: to turn on the virtual interface
tagged Y : it adds to the port Y the support of tagged frames for the current VLAN


socklab udp – it executes socklab with udp protocol
sock – it creates a udp socket, it’s equivalent to type sock udp and bind
sendto <Socket ID> <hostname> <port #> – emission of data packets
recvfrom <Socket ID> <byte #> – it receives data from socket
socklab tcp – it executes socklab with tcp protocol
passive – it creates a socket in passive mode, it’s equivalent to socklab, sock tcp, bind, listen
accept – it accepts an incoming connection (it can be done before or after creating the incoming connection)
connect <hostname> <port #> – these two commands are equivalent to socklab, sock tcp, bind, connect
close – it closes the connection
read <byte #> – to read bytes on the socket
write (ex. write ciao, ex. write #10) to write “ciao” or to write 10 bytes on the socket


rm /etc/resolv.conf – it prevent address resolution and make sure your filtering and firewall rules works properly
ipnat –f file_name – it writes filtering rules into file_name
ipnat –l – it gives the list of active rules
ipnat –C –F – it re-initialize the rules table
map em0 -> em0 : mapping IP addresses to the interface
map em0 -> portmap tcp/udp 20000:50000 : mapping with port
ipf –f file_name : it writes filtering rules into file_name
ipf –F –a : it resets the rule table
ipfstat –I : it grants access to a few information on filtered packets, as well as active filtering rules

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.