MTU/MSS Decision Tree for WireGuard Deployments

January 22, 2026 • Defguard Team

6 min to read

Overview

This guide covers common MTU/MSS scenarios for WireGuard. For complex network topologies or carrier-specific issues, consult your network engineering team.

Purpose:

This is a complete troubleshooting guide focused on solving MTU (Maximum Transmission Unit) and MSS (Maximum Segment Size) problems in WireGuard VPN setups. These issues often cause symptoms like stalled large file transfers, partial web page loads, SSH working but SCP failing, or intermittent connectivity — even when the VPN handshake succeeds.

For very complex networks or carrier-specific quirks, involve your network engineering team.

Quick Reference

Provides safe starting MTU values depending on the user’s internet connection type:

Path Type	Recommended MTU	Notes
Standard internet	1420	Safe default for most deployments
PPPoE (DSL/Fiber)	1412	PPPoE adds 8-byte overhead
Mobile/LTE	1400	Conservative for variable paths
Satellite	1400	High latency, avoid fragmentation
Data center (jumbo frames)	1420	WireGuard overhead still applies
Double NAT	1400	Extra encapsulation possible

Decision Tree

The guide gives a logical flowchart to quickly determine whether your issue is MTU-related:

START: User reports connectivity issues
    │
    ▼
┌─────────────────────────────────────────┐
│ Can user establish handshake?           │
│ Check: wg show wg0 latest-handshakes    │
└─────────────────────────────────────────┘
    │
    ├── NO ──► Handshake Issues (not MTU)
    │          - Check firewall (UDP 51820)
    │          - Check endpoint resolution
    │          - Check key configuration
    │
    ▼ YES
┌─────────────────────────────────────────┐
│ Can user ping gateway (small packets)?  │
│ ping -c 3 -s 64 <gateway_ip>            │
└─────────────────────────────────────────┘
    │
    ├── NO ──► Routing Issues (not MTU)
    │          - Check AllowedIPs
    │          - Check server-side routing
    │
    ▼ YES
┌─────────────────────────────────────────┐
│ Do large transfers fail or stall?       │
│ - Web pages partially load              │
│ - SSH works, SCP fails                  │
│ - Small pings OK, large pings timeout   │
└─────────────────────────────────────────┘
    │
    ├── NO ──► Not an MTU issue
    │          - Check bandwidth limits
    │          - Check application issues
    │
    ▼ YES
┌─────────────────────────────────────────┐
│ ══════════════════════════════════════  │
│         LIKELY MTU PROBLEM              │
│ ══════════════════════════════════════  │
└─────────────────────────────────────────┘
    │
    ▼
Proceed to MTU Discovery below

MTU Discovery Procedure

Step-by-step process to empirically find the maximum safe packet size through the tunnel.

Step 1: Find Working MTU (Client Side)

Test from client through VPN tunnel to a host on the other side:

# Linux/macOS - Start high, decrease until success
# -M do = don't fragment (Linux)
# -D = don't fragment (macOS)

# Linux - Start with 1472 (1500 - 28), decrease by 10s until success
ping -c 3 -M do -s 1472 <target_ip>

# macOS - Start with 1472
ping -c 3 -D -s 1472 <target_ip>

# Windows - Start with 1472
ping -n 3 -f -l 1472 <target_ip>

Step 2: Binary Search for Optimal MTU

# If 1472 fails, try progressively smaller values
ping -c 3 -M do -s 1462 <target_ip>  # Try 10 bytes smaller

# If that works, try larger values to find the boundary
ping -c 3 -M do -s 1468 <target_ip>  # Try 6 bytes larger

# Continue testing in 2-4 byte increments until you find the exact boundary

Step 3: Calculate Interface MTU

Interface MTU = Successful ping size + 28 (IP + ICMP headers)

Example: ping -s 1372 works
Interface MTU = 1372 + 28 = 1400

MTU Overhead Breakdown

Shows exactly where the overhead comes from in a WireGuard packet:

Outer: IP (20 IPv4 / 40 IPv6) + UDP (8) = 28 / 48 bytes
WireGuard: Header (32) + Authentication tag (16) = 48 bytes
Total overhead: 60 bytes (IPv4) or 80 bytes (IPv6)

┌─────────────────────────────────────────────────────────┐
│ Standard Ethernet Frame (1500 bytes)                    │
├─────────────────────────────────────────────────────────┤
│ ┌─────────────────────────────────────────────────────┐ │
│ │ Outer IP Header (20 bytes)                          │ │
│ ├─────────────────────────────────────────────────────┤ │
│ │ UDP Header (8 bytes)                                │ │
│ ├─────────────────────────────────────────────────────┤ │
│ │ WireGuard Header (32 bytes)                         │ │
│ ├─────────────────────────────────────────────────────┤ │
│ │ ┌─────────────────────────────────────────────────┐ │ │
│ │ │ Inner Packet (max 1440 bytes)                   │ │ │
│ │ │   - Inner IP Header (20 bytes)                  │ │ │
│ │ │   - Inner TCP Header (20 bytes)                 │ │ │
│ │ │   - Payload (max 1400 bytes)                    │ │ │
│ │ └─────────────────────────────────────────────────┘ │ │
│ ├─────────────────────────────────────────────────────┤ │
│ │ WireGuard Auth Tag (16 bytes)                       │ │
│ └─────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘

WireGuard overhead: 20 + 8 + 32 + 16 = 60 bytes (IPv4)
                    40 + 8 + 32 + 16 = 80 bytes (IPv6)

Safe WireGuard MTU: 1500 - 80 = 1420 (conservative, works for both IPv4 and IPv6)

Configuration Examples

WireGuard Interface MTU

Set MTU directly in WireGuard config (recommended):

# /etc/wireguard/wg0.conf
[Interface]
PrivateKey = <key>
Address = 10.0.0.2/24
MTU = 1400  # Conservative setting

[Peer]
PublicKey = <key>
Endpoint = vpn.example.com:51820
AllowedIPs = 0.0.0.0/0

Linux: Set MTU on Running Interface

Change MTU on a running interface (temporary):

# Temporary (until restart)
ip link set wg0 mtu 1400

# Verify
ip link show wg0 | grep mtu

MSS Clamping(alternative when you can’t/won’t change client MTU):

On the server, use iptables to clamp TCP MSS during handshake:
Automatic (best): bash--clamp-mss-to-pmtu
Manual example (for MTU 1420): bash--set-mss 1380 (1420 – 40 for TCP+IP headers)

When you cannot change client MTU, clamp MSS on the server to prevent TCP segmentation issues:

# On WireGuard server - Option 1: Automatic clamping to PMTU
iptables -t mangle -A FORWARD -i wg0 -p tcp \
  --tcp-flags SYN,RST SYN -j TCPMSS --clamp-mss-to-pmtu

# Option 2: Set explicit MSS (WireGuard MTU - 40 bytes for TCP/IP headers)
# For standard IPv4 WireGuard MTU 1420: MSS = 1420 - 40 = 1380
iptables -t mangle -A FORWARD -i wg0 -p tcp \
  --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1380

Common Scenarios

Scenario 1: PPPoE Connection

Problem: User on DSL/Fiber with PPPoE cannot load large pages Cause: PPPoE adds 8-byte overhead, reducing path MTU to 1492

Solution:

Set WireGuard MTU to 1412 (1492 - 80)
Or use MSS clamping to 1372 (1412 - 40)

Scenario 2: Mobile User

Problem: Laptop user reports intermittent failures on mobile hotspot Cause: Mobile carriers often have variable/reduced MTU

Solution:

Set WireGuard MTU to 1280 (minimum IPv6 MTU, always works)
Or use 1400 as reasonable compromise

Scenario 3: Double NAT / CGNAT

Problem: User behind carrier-grade NAT has connectivity issues Cause: Additional encapsulation in carrier network

Solution:

Reduce MTU to 1380 or lower
Test empirically with ping method above

Scenario 4: Corporate Firewall

Problem: User behind corporate proxy/firewall drops large packets (Cause: Firewall may filter or fragment oversized UDP).

Solution:

Test MTU from inside corporate network
May need MTU as low as 1280
Consider if firewall blocks UDP entirely (use TCP fallback if available)

Troubleshooting Commands

Check Current MTU

# Linux
ip link show wg0 | grep mtu

# macOS
ifconfig utun3 | grep mtu  # utun number varies

# Windows (PowerShell)
Get-NetIPInterface | Where-Object {$_.InterfaceAlias -like "*WireGuard*"}

Check Path MTU

# Linux - trace with MTU discovery
tracepath <destination>

# Check for ICMP fragmentation needed messages
tcpdump -i eth0 'icmp[0] == 3 and icmp[1] == 4'

Verify MSS Clamping

# Capture TCP SYN to see MSS option
tcpdump -i wg0 'tcp[tcpflags] & tcp-syn != 0' -v

Best Practices

Start conservative: Use 1400 or lower initially, optimize later
Test from affected client: MTU issues are path-specific
Document per-location settings: Different offices may need different MTU
Consider MSS clamping: Often easier than changing client configs
Monitor for regressions: ISP changes can alter path MTU

Quick Fix Cheat Sheet

Symptom	Quick Fix	Permanent Fix
Large transfers stall	Lower MTU to 1280	Find optimal MTU with ping test
Random disconnects	Lower MTU to 1400	Test path MTU, set precisely
Works on some networks	Set MTU to 1280	Create per-network profiles
VoIP/video issues	Check MTU + QoS	Ensure MTU consistent both ends

Core takeaway:

Most WireGuard “connectivity but no large traffic” problems are MTU-related. Use the decision tree first, then discover the real path MTU with ping tests, set a safe value (often 1400 or lower), or apply server-side MSS clamping as a simpler fix.