Find out if your network blocks WireGuard
WireGuard uses UDP and does not hide what it is. Some networks drop it on purpose, and some drop it by accident. This guide shows how to tell the difference between a blocked tunnel and a broken configuration, and what to do next.
The device sends a handshake
WireGuard sends a handshake message in one UDP packet to the address and port of the peer.
The firewall drops the packet
The firewall does not permit that UDP port, or it recognizes the handshake. It sends no error back.
The web continues to work
Web traffic on TCP port 443 passes, so the network looks healthy to all other apps.
The tunnel stays silent
The device tries again and gets no answer. It sends bytes and receives none.
Why a network blocks WireGuard
WireGuard sends all its traffic in UDP packets. It has no mode for TCP, and it does not try to look like a different protocol. These two facts explain almost every block.
The network filters UDP
Many guest, hotel, airport, and corporate networks permit only a short list of outbound traffic. The list usually has web traffic on TCP ports 80 and 443, and DNS. All other ports are closed, and that includes the UDP port of your tunnel. The network owner did not think about WireGuard at all. This is the most common cause.
Deep packet inspection recognizes the protocol
Deep packet inspection (DPI) looks inside packets, not only at port numbers. WireGuard is easy to recognize. A handshake starts with a message that has a fixed type and a fixed size of 148 bytes. The WireGuard protocol description gives the format. A firewall that looks for this pattern can block WireGuard on any port.
A policy or a law forbids VPN protocols
Some employers do not permit personal VPNs on the company network. Some countries filter VPN protocols at the national level. In these cases, the block is intentional.
This guide does not explain how to evade a block that an employer or a government made on purpose. The correct fix is to ask the network owner to allow the traffic.
The symptoms of a blocked tunnel
A firewall that blocks WireGuard usually drops packets and sends no error. WireGuard also stays silent when a packet is not valid. The result is that nothing fails loudly. Look for these signs.
- The handshake never completes. The client sends a handshake again and again, and no reply arrives.
- Zero bytes received. The transfer counter shows bytes sent, but the count of received bytes stays at zero.
- It works on mobile data but not on Wi-Fi. The same device and the same configuration connect through a phone hotspot.
- It worked yesterday on a different network. Nothing changed on the device, only the location.
- The tunnel connects but stops after a short time. Some firewalls permit the first packets and then cut long UDP flows.
"WireGuard not working" is also a common result of a wrong key, a wrong endpoint, or a server firewall. These problems give the same silent failure, and they fail on every network. A block fails on one network only.
Three tests to run
Run the tests in this order. Each one removes a group of possible causes.
1. Try another network
Connect the device to a phone hotspot and start the tunnel again. If the tunnel works, the configuration is correct and the first network is the cause. If the tunnel fails on the two networks, look at the configuration and at the server.
2. Check the latest handshake
On a device that uses the standard WireGuard tools, run wg show. Read two lines for the peer.
latest handshakeshows when the last WireGuard handshake succeeded. If the line is absent, no handshake completed.transfershows bytes received and bytes sent. Bytes sent with zero bytes received means that the replies do not arrive.
A handshake that is less than about two minutes old means that packets pass in the two directions. The problem is then inside the tunnel, such as routes, DNS, or MTU.
3. Check UDP egress
Test if the network permits outbound UDP at all. Start a video call or resolve a name with a public resolver, which uses UDP port 53. If other UDP apps work, the network blocks your port or recognizes the protocol. If no UDP app works, the network blocks UDP in general.
Find the cause
Answer the questions to narrow the cause. Each result tells you what to do next.
What to do about a block
The correct action depends on who owns the network.
- You own the network. Add an outbound rule for the UDP traffic of the tunnel. A stateful firewall then permits the replies. A stateless firewall needs a second rule for the UDP return traffic.
- Your employer owns the network. Send the IT team the exact hosts, ports, and protocols. A request that names specific ports gets approved much faster than a request to "allow the VPN".
- A hotel, a customer, or a public place owns the network. Ask, and use a different network if the answer is no. A phone hotspot is the usual alternative.
- A country filters VPN protocols. Follow the local law. This guide gives no method to get around the filter.
Some VPN products send traffic inside TLS on TCP port 443, which looks like web traffic to a simple firewall. Such a product passes more firewalls than WireGuard does. It also pays a price in speed, because it puts TCP inside TCP. If you must work from networks that permit only web traffic, a TLS-based VPN is the practical choice.
What Firezone does, and what it does not do
Firezone is zero trust access built on WireGuard, so the facts above apply to it.
What Firezone does not do
- Firezone sends WireGuard traffic over UDP. It does not hide or obfuscate the traffic.
- Firezone has no fallback to TCP port 443. On a network that blocks all outbound UDP, a Client cannot reach Resources.
What Firezone does
- No inbound port. Clients and Gateways make outbound connections only. In most cases, there is no inbound port to open and none for a firewall to block.
- Hole punching. A Client and a Gateway connect directly through NAT when the networks permit it.
- Relays. When a direct path fails, encrypted traffic goes through Firezone relays. The relays use UDP port 3478 and UDP ports 49152-65535.
- A published list. The ports reference gives each host, port, and protocol that a Client or Gateway needs. You can send it to a network owner as it is.
The control plane uses HTTPS on port 443. On a network that permits only web traffic, a Client can therefore sign in and still pass no traffic. If you see that pattern, check UDP egress first.
The wg show command does not apply to Firezone, because the Client does not use the standard WireGuard tools. Read the Client logs as an alternative. The troubleshooting guide gives the log locations.
Raw WireGuard, a TLS-based VPN, and Firezone on a strict network
The three approaches behave differently on a network that filters outbound traffic. No approach wins every row.
| Raw WireGuard | TLS-based VPN | Firezone | |
|---|---|---|---|
| Transport | UDP | TCP 443, sometimes UDP | UDP |
| Works when the network blocks all UDP | No | YesLooks like web traffic | No |
| Hard to recognize with deep packet inspection | NoFixed handshake format | PartialDepends on the product | NoDoes not obfuscate traffic |
| No inbound port on the server side | NoOne open UDP port | NoOne open TCP port | YesOutbound connections only |
| Connects through NAT on the two sides | NoOne side needs a reachable port | NoThe server needs a public address | YesHole punching, then relays |
| Avoids TCP inside TCP | Yes | PartialNot on TCP 443 | Yes |
| Published list of ports to allow | One port that you select | Depends on the product | YesIn the ports reference |
The table describes the usual design of each category. A specific product can be different. Read the documentation of the product before you decide.
Common questions
How do I know if my network blocks WireGuard?
Start the tunnel on the network, then start it on a phone hotspot. If it works only on the hotspot, the first network blocks it. On the network that fails, wg show shows no latest handshake and zero bytes received.
Why does WireGuard work on mobile data but not on Wi-Fi?
The Wi-Fi network filters outbound UDP, or it blocks the port of your tunnel. Mobile networks usually permit outbound UDP. The configuration is correct, because it works on one of the two networks.
Can a firewall block WireGuard on any port?
Yes. A firewall with deep packet inspection can recognize the first handshake message by its fixed type and size. A port change does not help against such a firewall.
Does WireGuard work over TCP port 443?
No. WireGuard uses UDP only. A tool that wraps WireGuard in TCP is a separate product with its own cost in speed. Firezone does not include such a wrapper.
Which ports does Firezone need?
Clients and Gateways need outbound HTTPS on port 443 to the control plane. They also need UDP port 3478 for STUN, UDP ports 49152-65535 for the relays, and UDP for WireGuard traffic. The ports reference has the full list of hosts.
Is UDP blocked on most corporate networks?
Many corporate networks limit outbound UDP to DNS and a few approved services. Guest networks are frequently stricter. The fix is a firewall rule, and the IT team needs the exact ports to write it.