veth(4)

✨ veth — Virtual Ethernet Device ✨

okay bestie, gather round, because this one is genuinely iconic 💖 it's the veth page from the kernel interfaces manual, and it's giving main character energy in every container you've ever touched.

---

🎀 NAME 🎀

veth — Virtual Ethernet Device

---

🌸 DESCRIPTION 🌸

So. veth devices are virtual Ethernet devices. That's the whole vibe. They can act as tunnels between network namespaces — like, you can use one to create a bridge to a physical network device that's chillin' in another namespace — but ngl they also work perfectly fine as standalone network devices all on their own. versatile queen 💅

Here's the thing though, and this is important: veth devices are always created in interconnected pairs. no solo acts, ever. it's a duo situation, they come as a set 🫶

You make a pair like this:

# ip link add <p1-name> type veth peer name <p2-name>

In that command, p1-name and p2-name are the names you're assigning to the two connected end points. cute, right?

Now the actual behavior: packets transmitted on one device in the pair are immediately received on the other device. instant. no latency drama. BUT — and this is the catch — when either device is down, the link state of the pair is down. both of them. they're linked like that, it's a whole thing 😭

---

🦋 why you'd even want this

veth device pairs are useful for combining the network facilities of the kernel together in interesting ways. that's the official phrasing and honestly? slay.

A particularly interesting use case — and this is the one everybody's obsessed with — is to place one end of a veth pair in one network namespace and the other end in another network namespace. that way you get communication between network namespaces. it's giving tunnel but make it namespace-y ✨

To do that, you can just hand over the netns parameter when you're creating the interfaces:

# ip link add <p1-name> netns <p1-ns> type veth peer <p2-name> netns <p2-ns>

or, if you've already got an existing veth pair and you're like "hmm actually let me move one side", you move one side to the other namespace:

# ip link set <p2-name> netns <p2-ns>

periodt. 💅

---

💜 finding your other half

okay so you've got a veth interface and you're like "who is your peer bestie" — that's where ethtool(8) comes in. you can use it to find the peer of a veth network interface, with commands something like:

# ip link add ve_A type veth peer name ve_B;   # Create veth pair
# ethtool -S ve_A;         # Discover interface index of peer
NIC statistics:
     peer_ifindex: 16
# ip link | grep '^16:';   # Look up interface
16: ve_B@ve_A: <BROADCAST,MULTICAST,M-DOWN> mtu 1500 qdisc ...

see that? peer_ifindex: 16 told you where to look, and then ip link | grep '^16:' showed you it's ve_B@ve_A, which is literally the other half of your pair. mystery solved, friendship confirmed 🥹💖

---

🌈 SEE ALSO 🌈

• clone(2)
• network_namespaces(7)
• ip(8)
• ip-link(8)
• ip-netns(8)

---

and that's the whole page bestie, you ate it 🍓✨

this came from Linux man-pages 6.20, dated 2026-02-08, and the manual name is veth in section 4 (kernel interfaces, for the record). go forth and pair your devices, ily 💜👑