ifconfig: the complete reference

ifconfig (interface configuration) is the original Unix tool for viewing and setting network interface parameters. It first appeared in 4.2BSD in 1983 and still ships on macOS and every BSD. On Linux it belongs to the net-tools package, is deprecated in favour of ip, and is often not installed. This page covers all three families, because the output and the syntax differ more than most guides admit.

Quick reference

Which ifconfig do you have?

There are at least four independent implementations, which explains why syntax that works on one machine fails on another.

Reading the output on Linux

net-tools 2.x (Ubuntu 18.04 and later, Debian 10 and later, RHEL 8 and later) prints this layout:

ifconfig eth0

eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 192.168.1.42  netmask 255.255.255.0  broadcast 192.168.1.255
        inet6 fe80::a00:27ff:fe4e:66a1  prefixlen 64  scopeid 0x20<link>
        inet6 2a00:23c4:5f1a:8b00:a00:27ff:fe4e:66a1  prefixlen 64  scopeid 0x0<global>
        ether 08:00:27:4e:66:a1  txqueuelen 1000  (Ethernet)
        RX packets 184213  bytes 231770215 (221.0 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 97312  bytes 11284570 (10.7 MiB)
        TX errors 0  dropped 0  overruns 0  carrier 0  collisions 0

eth0:

The interface name. Modern systemd distributions use predictable names such as enp3s0, ens192, eno1 or wlp2s0 instead of eth0 and wlan0. How the names are built →

flags=4163<UP,BROADCAST,RUNNING,MULTICAST>

The interface flags as a decimal number and as names. 4163 = 0x1043 = UP (0x1) + BROADCAST (0x2) + RUNNING (0x40) + MULTICAST (0x1000). See the flag table. UP without RUNNING means the interface is enabled but has no link: cable unplugged or Wi-Fi not associated.

mtu 1500

Maximum transmission unit in bytes. 1500 is standard Ethernet; 9000 is jumbo frames; 1280 is the IPv6 minimum; PPPoE links are often 1492 and VPN tunnels 1400 or lower.

inet / netmask / broadcast

IPv4 address, mask and broadcast address. Old net-tools shows only one IPv4 address per interface even if the kernel has several; that is one of the reasons it was replaced.

inet6 … prefixlen 64 scopeid 0x20<link>

IPv6 addresses with their prefix length and scope: <link> for fe80:: link-local, <global> for routable addresses, <host> for ::1, <site> for the deprecated site-local range.

ether 08:00:27:4e:66:a1

The MAC address. 08:00:27 is the VirtualBox prefix, so this is a VM. Followed by the transmit queue length and the link type in brackets.

RX / TX packets, bytes

Counters since boot. errors are frames with checksum or framing problems (bad cable, duplex mismatch); dropped are packets the kernel discarded because a buffer was full; overruns mean the NIC ring buffer filled before the driver could empty it; frame counts misaligned frames; carrier counts link losses during transmit; collisions should be zero on any switched network.

The older Linux layout

net-tools 1.60 (CentOS 6 and 7, Ubuntu 16.04, most tutorials written before 2018) prints the same information in a different shape. If you are following an old tutorial, this is the output it shows:

eth0      Link encap:Ethernet  HWaddr 08:00:27:4e:66:a1
          inet addr:192.168.1.42  Bcast:192.168.1.255  Mask:255.255.255.0
          inet6 addr: fe80::a00:27ff:fe4e:66a1/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:184213 errors:0 dropped:0 overruns:0 frame:0
          TX packets:97312 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:231770215 (221.0 MiB)  TX bytes:11284570 (10.7 MiB)

Short form

ifconfig -s

Iface      MTU    RX-OK RX-ERR RX-DRP RX-OVR    TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0      1500   184213      0      0 0         97312      0      0      0 BMRU
lo       65536     2210      0      0 0          2210      0      0      0 LRU

The Flg column abbreviates the flags: B broadcast, M multicast, R running, U up, L loopback, P point-to-point, O no ARP, A allmulti, D debug, m master, s slave.

Reading the output on macOS

macOS prints one block per interface, and there are many interfaces. On a MacBook you will typically see lo0, gif0, stf0, anpi0, en0 (Wi-Fi), en1 to en5 (Thunderbolt, USB Ethernet, bridges), ap1, awdl0 (AirDrop), llw0, bridge0 and utun0 to utun4 (VPN and iCloud Private Relay tunnels). The one you want is usually en0.

ifconfig en0

en0: flags=8863<UP,BROADCAST,SMART,RUNNING,SIMPLEX,MULTICAST> mtu 1500
	options=6460<TSO4,TSO6,CHANNEL_IO,PARTIAL_CSUM,ZEROINVERT_CSUM>
	ether f0:18:98:3a:2b:1c
	inet6 fe80::1c2b:3d4e:5f60:7a8b%en0 prefixlen 64 secured scopeid 0xc
	inet 192.168.1.23 netmask 0xffffff00 broadcast 192.168.1.255
	inet6 2a00:23c4:5f1a:8b00:8c1:2f3e:4d5a:6b7c prefixlen 64 autoconf secured
	inet6 2a00:23c4:5f1a:8b00:f1e2:d3c4:b5a6:9788 prefixlen 64 autoconf temporary
	nd6 options=201<PERFORMNUD,DAD>
	media: autoselect
	status: active

netmask 0xffffff00

BSD prints masks in hexadecimal. 0xffffff00 is 255.255.255.0 (/24); 0xffff0000 is 255.255.0.0 (/16); 0xfffffe00 is 255.255.254.0 (/23). Convert with the mask table.

options=…

Hardware offload capabilities (TCP segmentation offload, checksum offload). Informational.

secured / autoconf / temporary

macOS marks IPv6 addresses by origin: autoconf from SLAAC, secured means a stable privacy address generated per RFC 7217 (not derived from the MAC), temporary is a rotating privacy address, dynamic comes from DHCPv6.

media: autoselect / status: active

Link negotiation and link state. status: inactive means no link. On Wi-Fi you may see media: autoselect (<unknown type>); that is normal.

SIMPLEX, SMART

BSD-specific flags. SIMPLEX means the hardware cannot hear its own transmissions; SMART is an Apple driver flag. Neither needs attention.

Useful macOS-only switches: ifconfig -l lists just the names on one line; ifconfig -u shows only interfaces that are up; ifconfig -v en0 adds link quality and per-interface statistics; ifconfig -L en0 shows IPv6 address lifetimes; ifconfig -m en0 lists the media types the hardware supports.

Reading the output on FreeBSD and OpenBSD

ifconfig em0

em0: flags=1008843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST,LOWER_UP> metric 0 mtu 1500
	options=4e524bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCSUM,LRO,WOL_MAGIC,VLAN_HWFILTER,NOMAP,TXTLS4,TXTLS6,VXLAN_HWCSUM,VXLAN_HWTSO,TXTLS_RTLMT>
	ether 00:0c:29:5b:81:f3
	inet 192.168.1.60 netmask 0xffffff00 broadcast 192.168.1.255
	inet6 fe80::20c:29ff:fe5b:81f3%em0 prefixlen 64 scopeid 0x1
	media: Ethernet autoselect (1000baseT <full-duplex>)
	status: active
	nd6 options=23<PERFORMNUD,ACCEPT_RTADV,AUTO_LINKLOCAL>

FreeBSD 14 prints LOWER_UP when the physical link is up, and the media line tells you negotiated speed and duplex, which is what Linux users get from ethtool. OpenBSD adds index, priority, llprio and, for DHCP-managed interfaces, an inet autoconf note. Interface names follow the driver: em0, igb0, re0, vtnet0, ix0, wlan0, bge0.

Interface flags

Configuring interfaces

All of these need root: prefix with sudo. Note the syntax differences between Linux and BSD/macOS, which are the single most common cause of "ifconfig doesn't work" complaints.

Bring an interface up or down

sudo ifconfig eth0 down
sudo ifconfig eth0 up

On macOS use en0; downing Wi-Fi this way disconnects it and it will reconnect when brought back up. On Linux with NetworkManager, downing an interface may trigger NetworkManager to bring it back up seconds later; use nmcli device disconnect eth0 instead.

Set an IPv4 address

sudo ifconfig eth0 192.168.1.50 netmask 255.255.255.0
# broadcast is derived automatically; set it explicitly if needed
sudo ifconfig eth0 192.168.1.50 netmask 255.255.255.0 broadcast 192.168.1.255
# CIDR form (net-tools 2.x only)
sudo ifconfig eth0 192.168.1.50/24

sudo ifconfig en0 inet 192.168.1.50 netmask 255.255.255.0
# the "inet" keyword is required on BSD-derived systems

sudo ifconfig em0 inet 192.168.1.50/24
sudo ifconfig em0 inet 192.168.1.50 netmask 255.255.255.0

Add a second address (alias)

# classic alias interface (shows up as eth0:1)
sudo ifconfig eth0:1 192.168.1.51 netmask 255.255.255.0 up
# remove it
sudo ifconfig eth0:1 down
# net-tools 2.x also has add/del for IPv6 and IPv4
sudo ifconfig eth0 add 2001:db8::50/64
sudo ifconfig eth0 del 2001:db8::50/64

sudo ifconfig en0 alias 192.168.1.51 netmask 255.255.255.255
sudo ifconfig en0 -alias 192.168.1.51
# IPv6
sudo ifconfig en0 inet6 2001:db8::50 prefixlen 64 alias
sudo ifconfig en0 inet6 2001:db8::50 -alias

Set the MTU

sudo ifconfig eth0 mtu 9000     # Linux
sudo ifconfig en0 mtu 1400      # macOS/BSD, same syntax

Jumbo frames only help if every device on the path supports them. A too-large MTU on a VPN or PPPoE link causes stalled downloads and pages that half-load; try 1400 if that describes your problem.

Change the MAC address

sudo ifconfig eth0 down
sudo ifconfig eth0 hw ether 02:11:22:33:44:55
sudo ifconfig eth0 up

sudo ifconfig en0 ether 02:11:22:33:44:55

sudo ifconfig em0 ether 02:11:22:33:44:55

Promiscuous mode

sudo ifconfig eth0 promisc
sudo ifconfig eth0 -promisc

Capture tools set this for you. If you see PROMISC in the flags and are not running a capture, find out why: ip link and the kernel log (dmesg | grep promisc) record who enabled it.

Other keywords (Linux net-tools)

Why ifconfig was deprecated on Linux

ifconfig on Linux talks to the kernel through ioctl calls designed in the 1980s. The Linux networking stack moved on, and the old interface simply cannot express what the kernel can do. Concretely, net-tools ifconfig:

• Shows only the first IPv4 address on an interface, silently hiding the rest.
• Cannot manage policy routing, multiple routing tables, or route metrics properly.
• Knows nothing about network namespaces, VLANs, bridges, bonds, VRFs, VXLAN, MACVLAN, tun/tap, or traffic control.
• Cannot show or set IPv6 address lifetimes, flags or the neighbour cache.
• Cannot produce machine-readable output.
• Prints counters using 32-bit-era formatting that wraps on busy interfaces in some versions.

Alexey Kuznetsov wrote iproute2 in 1999 to expose the new rtnetlink interface. Distributions kept net-tools for years for compatibility, then stopped installing it by default: Arch in 2011, Debian and Ubuntu around 2018, RHEL and CentOS with version 7 (2014), Fedora with 18. Debian's net-tools maintainers wrote in 2009 that the tools "should be considered obsolete". The package still exists and still works for simple tasks; it just is not the tool the kernel developers or the distributions maintain going forward. Full history →

None of this applies to BSD. FreeBSD's ifconfig is actively developed and is the right tool there; it gained JSON-ish output via -f format flags, VLAN, bridge, lagg, Wi-Fi scanning and epair support over the years.

ifconfig to ip translation

The full ip command reference goes much further.

Common errors

Installing ifconfig

sudo apt update && sudo apt install net-tools

sudo dnf install net-tools

sudo pacman -S net-tools

# BusyBox already provides a minimal ifconfig; for the full version:
sudo apk add net-tools

sudo zypper install net-tools-deprecated

openSUSE's package name says it plainly. If you are installing it only out of habit, spend the same five minutes learning ip -br addr, ip route and ip neigh; they cover everything ifconfig, route and arp did.

Related pages

• ip command reference, the replacement on Linux
• ifconfig: command not found
• macOS networking commands
• FreeBSD and OpenBSD networking
• ipconfig vs ifconfig
• History: 4.2BSD to iproute2