STUNMESH is a Wireguard helper tool to get through Full-Cone NAT.
Inspired by manuels' wireguard-p2p project
Supported Platform:
- Linux (amd64, arm, arm64, mips)
- MacOS* (amd64, arm64)
- FreeBSD (amd64, arm64)
Note
We only support wireguard-go in MacOS, Wireguard App store version is not supported because of sandbox currently.
Tested with
- VyOS 2025.07.14-0022-rolling (built-in Wireguard kernel module)
- Ubuntu with Wireguard in Kernel module
- MacOS Wireguard-go 0.0.20230223, Wireguard-tools 1.0.20210914
- FreeBSD 14.3-RELEASE (built-in Wireguard)
- OPNSense 25.1 (built-in Wireguard)
Use raw socket and cBPF filter to send and receive STUN 5389's packet to get public ip and port with same port of wireguard interface.
Encrypt public info with Curve25519 sealedbox and save it using configured storage plugins.
stunmesh-go will create and update records with domain <sha1 in hex>.<subdomain>.<your_domain>
(or <sha1 in hex>.<your_domain>
if no subdomain configured).
Once getting info from internet, it will setup peer endpoint with wireguard tools.
✅ Plugin system supported - Multiple storage backends with flexible configuration - supports exec plugin for custom implementations
Linux: Uses raw sockets with BPF filtering to listen on all interfaces system-wide. No interface-specific limitations.
FreeBSD and macOS (BSD-based systems): Uses BPF with interface-specific packet capture. stunmesh-go will listen on all eligible network interfaces for STUN response messages, excluding the specific WireGuard interface being managed. This provides better resilience for systems with multiple network paths or backup routes compared to single default route dependency.
make all
Note
For FreeBSD and MacOS, please use GNU Makefile gmake
to build.
sudo ./stunmesh-go
Or use STUNMESH-go in the container
docker pull tjjh89017/stunmesh
Put the configuration below paths:
/etc/stunmesh/config.yaml
~/.stunmesh/config.yaml
./config.yaml
---
refresh_interval: "1m"
log:
level: "debug"
interfaces:
wg0:
peers:
"<PEER_NAME>":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: cloudflare1
ping:
enabled: true
target: "192.168.1.100"
wg1:
peers:
"<PEER_NAME>":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: cloudflare2
ping:
enabled: true
target: "10.0.0.50"
interval: "60s"
timeout: "10s"
"<PEER_NAME>":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: exec_plugin1
# ping configuration is completely optional
stun:
address: "stun.l.google.com:19302"
ping_monitor:
interval: "5s"
timeout: "2s"
fixed_retries: 3
plugins:
cloudflare1:
type: cloudflare
zone_name: "example.com"
subdomain: "wg"
api_token: "${CLOUDFLARE_API_TOKEN}"
cloudflare2:
type: cloudflare
zone_name: "example.com"
api_token: "${CLOUDFLARE2_API_TOKEN}"
exec_plugin1:
type: exec
command: "python3"
args: ["/path/to/script.py", "--config", "/path/to/config"]
stunmesh-go now supports a flexible plugin system that allows you to:
- Multiple storage backends: Use different storage solutions for different peers
- Named plugin instances: Configure multiple instances of the same plugin type
- Per-peer plugin assignment: Each peer can use a different plugin instance
Cloudflare DNS Plugin (type: cloudflare
)
- Stores peer information in Cloudflare DNS TXT records
- Configuration:
zone_name
: Your domain name managed by Cloudflaresubdomain
: Subdomain prefix for DNS records (optional, defaults to empty)api_token
: Cloudflare API token with DNS edit permissions
Exec Plugin (type: exec
)
- Executes external scripts/programs for storage operations
- Configuration:
command
: Command to executeargs
: Command line arguments (optional)
- Protocol: JSON over stdin/stdout
The exec plugin communicates with external programs using JSON over stdin/stdout. Your program should:
- Read JSON request from stdin
- Process the request (GET or SET operation)
- Write JSON response to stdout
- Exit with code 0 for success, non-zero for error
Request Format:
{
"operation": "get|set",
"key": "peer_identifier_string",
"value": "encrypted_data_for_set_operation"
}
Response Format:
{
"success": true,
"value": "encrypted_data_for_get_operation",
"error": "error_message_if_failed"
}
Python Example (/usr/local/bin/stunmesh-storage.py
):
#!/usr/bin/env python3
import json
import sys
import os
# Simple file-based storage
STORAGE_DIR = "/var/lib/stunmesh"
def ensure_storage_dir():
os.makedirs(STORAGE_DIR, exist_ok=True)
def get_value(key):
file_path = os.path.join(STORAGE_DIR, f"{key}.txt")
try:
with open(file_path, 'r') as f:
return f.read().strip()
except FileNotFoundError:
return None
def set_value(key, value):
ensure_storage_dir()
file_path = os.path.join(STORAGE_DIR, f"{key}.txt")
with open(file_path, 'w') as f:
f.write(value)
def main():
try:
# Read JSON request from stdin
request = json.load(sys.stdin)
operation = request.get("operation")
key = request.get("key")
if operation == "get":
value = get_value(key)
if value is not None:
response = {"success": True, "value": value}
else:
response = {"success": False, "error": "Key not found"}
elif operation == "set":
value = request.get("value")
set_value(key, value)
response = {"success": True}
else:
response = {"success": False, "error": "Unknown operation"}
# Write JSON response to stdout
json.dump(response, sys.stdout)
except Exception as e:
response = {"success": False, "error": str(e)}
json.dump(response, sys.stdout)
sys.exit(1)
if __name__ == "__main__":
main()
Bash Example (/usr/local/bin/stunmesh-storage.sh
):
#!/bin/bash
STORAGE_DIR="/var/lib/stunmesh"
mkdir -p "$STORAGE_DIR"
# Read JSON from stdin
INPUT=$(cat)
# Parse JSON using jq
OPERATION=$(echo "$INPUT" | jq -r '.operation')
KEY=$(echo "$INPUT" | jq -r '.key')
case "$OPERATION" in
"get")
FILE_PATH="$STORAGE_DIR/${KEY}.txt"
if [ -f "$FILE_PATH" ]; then
VALUE=$(cat "$FILE_PATH")
echo "{\"success\": true, \"value\": \"$VALUE\"}"
else
echo "{\"success\": false, \"error\": \"Key not found\"}"
fi
;;
"set")
VALUE=$(echo "$INPUT" | jq -r '.value')
FILE_PATH="$STORAGE_DIR/${KEY}.txt"
echo "$VALUE" > "$FILE_PATH"
echo "{\"success\": true}"
;;
*)
echo "{\"success\": false, \"error\": \"Unknown operation\"}"
exit 1
;;
esac
Redis Example (/usr/local/bin/stunmesh-redis.py
):
#!/usr/bin/env python3
import json
import sys
import redis
# Redis connection
r = redis.Redis(host='localhost', port=6379, db=0, decode_responses=True)
def main():
try:
request = json.load(sys.stdin)
operation = request.get("operation")
key = f"stunmesh:{request.get('key')}"
if operation == "get":
value = r.get(key)
if value is not None:
response = {"success": True, "value": value}
else:
response = {"success": False, "error": "Key not found"}
elif operation == "set":
value = request.get("value")
r.set(key, value)
# Optional: Set expiration (e.g., 24 hours)
r.expire(key, 86400)
response = {"success": True}
else:
response = {"success": False, "error": "Unknown operation"}
json.dump(response, sys.stdout)
except Exception as e:
response = {"success": False, "error": str(e)}
json.dump(response, sys.stdout)
sys.exit(1)
if __name__ == "__main__":
main()
Configuration Examples:
plugins:
file_storage:
type: exec
command: "/usr/local/bin/stunmesh-storage.py"
bash_storage:
type: exec
command: "/usr/local/bin/stunmesh-storage.sh"
redis_storage:
type: exec
command: "python3"
args: ["/usr/local/bin/stunmesh-redis.py"]
remote_api:
type: exec
command: "curl"
args: ["-s", "-X", "POST", "-H", "Content-Type: application/json", "--data-binary", "@-", "https://api.example.com/stunmesh"]
stunmesh-go supports intelligent ping monitoring to detect tunnel health and automatically trigger reconnection when issues are detected.
- Per-peer ping monitoring: Each peer can have its own target IP and ping settings
- Adaptive retry logic: Intelligent failure handling with exponential backoff
- Automatic recovery: Triggers publish/establish operations on ping failures
- Configurable timeouts: Per-peer or global timeout and interval settings
- Normal Operation: Pings target IP at configured intervals (constant frequency)
- Failure Detection: When ping fails, immediately triggers publish and establish for the specific failing peer
- Separate Retry Logic:
- Ping monitoring: Continues at constant configured interval regardless of failures
- Publish/Establish retries: Independent schedule with adaptive backoff
- All retries: Always publish endpoint for the specific failing peer, then establish connection
- First 3 retries: Fixed 2-second intervals
- After 3 retries: Arithmetic progression backoff (10s, 12s, 14s, 16s, 18s...)
- Cap at
refresh_interval
: Hands over to normal refresh cycle, ping monitoring continues
- Recovery: On successful ping, resets retry logic and re-enables publish/establish
Global Ping Settings:
ping_monitor:
interval: "5s" # Default ping interval
timeout: "2s" # Default ping timeout
fixed_retries: 3 # Fixed retry attempts before exponential backoff
Per-Peer Ping Configuration:
interfaces:
wg0:
peers:
"peer1":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: cloudflare1
ping:
enabled: true
target: "192.168.1.100" # IP to ping through tunnel
interval: "60s" # Override global interval (optional)
timeout: "10s" # Override global timeout (optional)
"peer2":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: cloudflare2
ping:
enabled: true
target: "10.0.0.50"
# Uses global interval and timeout defaults
"peer3":
public_key: "<PUBLIC_KEY_IN_BASE64>"
plugin: cloudflare2
# No ping configuration = ping monitoring disabled
Note: The entire ping
section is optional. If omitted, ping monitoring is disabled for that peer.
enabled
: Enable ping monitoring for this peer (default: false)target
: IP address to ping through the tunnel (required if enabled)interval
: How often to ping (optional, uses global default)timeout
: Max time to wait for ping response (optional, uses global default)
- IPv4 Only: Ping monitoring currently supports IPv4 addresses only
- IP Address Required: The
target
field must be an IP address, not a domain name - Examples:
- ✅ Valid:
"192.168.1.100"
,"10.0.0.1"
,"172.16.0.50"
- ❌ Invalid:
"router.local"
,"google.com"
,"2001:db8::1"
- ✅ Valid:
- Tunnel Health Monitoring: Detect when WireGuard tunnel stops working
- Automatic Recovery: Reconnect without manual intervention
- Network Redundancy: Faster failover than waiting for refresh_interval
- Proactive Maintenance: Identify and fix connectivity issues quickly
These are the strict examples to show you how to use STUNMESH-go in your environment. You can mix the setup in your envrionments.
For example, you can use VyOS router with Mac with STUNMESH-go to connect Wireguard tunnel with only Mobile networks or Public IPs.
Suppose you have two VyOS routers with LTE modem on them.
Hardware in Site A:
- VyOS_A with LTE modem and SIM card.
- Other devices under the subnet.
Hardware in Site B:
- VyOS_B with LTE modem and SIM card.
- Other devices under the subnet.
- Configure VyOS_A with LTE connections.
- Configure VyOS_A with the following commands.
- Wait for it connected. (Mostly, it will require 2 times of refresh_interval)
VyOS Commands
mkdir -p /config/user-data/stunmesh
cat <<EOF > /config/user-data/stunmesh/config.yaml
refresh_interval: "1m"
log:
level: "debug"
interfaces:
wg0:
peers:
"VYOS_B":
public_key: "<VYOS_B_PUBLIC_KEY>"
plugin: cloudflare_main
stun:
address: "stun.l.google.com:19302"
plugins:
cloudflare_main:
type: cloudflare
zone_name: "<ZONE_NAME>"
api_token: "<API_TOKEN>"
EOF
configure
set container name stunmesh allow-host-networks
set container name stunmesh capability 'net-admin'
set container name stunmesh capability 'net-raw'
set container name stunmesh capability 'net-bind-service'
set container name stunmesh capability 'sys-admin'
set container name stunmesh image 'tjjh89017/stunmesh'
set container name stunmesh uid '0'
set container name stunmesh volume certs destination '/etc/ssl/certs'
set container name stunmesh volume certs mode 'ro'
set container name stunmesh volume certs source '/etc/ssl/certs'
set container name stunmesh volume config destination '/etc/stunmesh'
set container name stunmesh volume config mode 'ro'
set container name stunmesh volume config source '/config/user-data/stunmesh'
set interfaces wireguard wg0 address '192.168.10.1/24'
set interfaces wireguard wg0 port '<YOUR_WIREGUARD_PORT>'
set interfaces wireguard wg0 ip adjust-mss '1380'
set interfaces wireguard wg0 ipv6 adjust-mss '1360'
set interfaces wireguard wg0 mtu '1420'
set interfaces wireguard wg0 peer VYOS_B allowed-ips '192.168.10.2/24'
set interfaces wireguard wg0 peer VYOS_B persistent-keepalive '15'
set interfaces wireguard wg0 peer VYOS_B public-key <VYOS_B_PUBLIC_KEY>
set interfaces wireguard wg0 private-key <VYOS_A_PRIVATE_KEY>
# You will need to setup firewall rules to allow ingress traffic to '<YOUR_WIREGUARD_PORT>'
# Please check the VyOS docs to use nft style firewall or Zone Based Firewall
commit
save
- Configure VyOS_B with LTE connections.
- Configure VyOS_B with the following commands.
- Wait for it connected. (Mostly, it will require 2 times of refresh_interval)
VyOS Commands
mkdir -p /config/user-data/stunmesh
cat <<EOF > /config/user-data/stunmesh/config.yaml
refresh_interval: "1m"
log:
level: "debug"
interfaces:
wg0:
peers:
"VYOS_A":
public_key: "<VYOS_A_PUBLIC_KEY>"
plugin: cloudflare_main
stun:
address: "stun.l.google.com:19302"
plugins:
cloudflare_main:
type: cloudflare
zone_name: "<ZONE_NAME>"
api_token: "<API_TOKEN>"
EOF
configure
set container name stunmesh allow-host-networks
set container name stunmesh capability 'net-admin'
set container name stunmesh capability 'net-raw'
set container name stunmesh capability 'net-bind-service'
set container name stunmesh capability 'sys-admin'
set container name stunmesh image 'tjjh89017/stunmesh'
set container name stunmesh uid '0'
set container name stunmesh volume certs destination '/etc/ssl/certs'
set container name stunmesh volume certs mode 'ro'
set container name stunmesh volume certs source '/etc/ssl/certs'
set container name stunmesh volume config destination '/etc/stunmesh'
set container name stunmesh volume config mode 'ro'
set container name stunmesh volume config source '/config/user-data/stunmesh'
set interfaces wireguard wg0 address '192.168.10.2/24'
set interfaces wireguard wg0 port '<YOUR_WIREGUARD_PORT>'
set interfaces wireguard wg0 ip adjust-mss '1380'
set interfaces wireguard wg0 ipv6 adjust-mss '1360'
set interfaces wireguard wg0 mtu '1420'
set interfaces wireguard wg0 peer VYOS_A allowed-ips '192.168.10.2/24'
set interfaces wireguard wg0 peer VYOS_A persistent-keepalive '15'
set interfaces wireguard wg0 peer VYOS_A public-key <VYOS_A_PUBLIC_KEY>
set interfaces wireguard wg0 private-key <VYOS_B_PRIVATE_KEY>
# You will need to setup firewall rules to allow ingress traffic to '<YOUR_WIREGUARD_PORT>'
# Please check the VyOS docs to use nft style firewall or Zone Based Firewall
commit
save
Ping each other with Wireguard interface's IP to test the connection
You may need to configure some static route or dynamic route to connect two subnets with different sites.
Suppose you have two LTE/5G routers and two downlink MacOS or Linux computers. Here we use the following setup to demo.
Hardware in Site A:
- Netgear M5 5G router (with Asia Pacific Telecom, Now Far EasTone Telecom)
- MacOS Intel-based
Intel Mac
Hardware in Site B:
- iPhone 15 Pro (with Chunghwa Telecom 4G LTE SIM)
- Macbook Air M3
Mac M3
- Connect your
Intel Mac
with Netgear M5 to get the internet. - Install wireguard-go and wireguard-tools in your Mac.
- Download STUNMESH-go for your Mac's architecture to
/tmp/stunmesh-go
. - Prepare your wireguard configuration as below
- Prepare
config.yaml
as below, please fill yourutunX
interface from the result ofwg-quick
- Run wireguard tunnel with
wg-quick up /tmp/wg0.conf
- Run STUNMESH-go in
/tmp/
,cd /tmp; sudo ./stunmesh-go
- Wait for it connected. (Mostly, it will require 2 times of
refresh_interval
)
Wiregaurd Config /tmp/wg0.conf
[Interface]
PrivateKey = <INTEL_MAC_PRIVATE_KEY>
Address = 192.168.10.1/24
[Peer]
PublicKey = <MAC_M3_PUBLIC_KEY>
AllowedIPs = 192.168.10.0/24
PersistentKeepalive = 25
STUNMESH-go /tmp/config.yaml
---
refresh_interval: "1m"
log:
level: "debug"
interfaces:
"<utunX>":
peers:
"MAC_M3":
public_key: "<MAC_M3_PUBLIC_KEY>"
plugin: cloudflare_main
stun:
address: "stun.l.google.com:19302"
plugins:
cloudflare_main:
type: cloudflare
zone_name: "<ZONE_NAME>"
api_token: "<API_TOKEN>"
- Connect your
Mac M3
with your iPhone15 to get the internet. - Install wireguard-go and wireguard-tools in your Mac.
- Download STUNMESH-go for your Mac's architecture to
/tmp/stunmesh-go
. - Prepare your wireguard configuration as below
- Prepare
config.yaml
as below, please fill yourutunX
interface from the result ofwg-quick
- Run wireguard tunnel with
wg-quick up /tmp/wg0.conf
- Run STUNMESH-go in
/tmp/
,cd /tmp; sudo ./stunmesh-go
- Wait for it connected. (Mostly, it will require 2 times of
refresh_interval
)
Wiregaurd Config /tmp/wg0.conf
[Interface]
PrivateKey = <MAC_M3_PRIVATE_KEY>
Address = 192.168.10.2/24
[Peer]
PublicKey = <INTEL_MAC_PUBLIC_KEY>
AllowedIPs = 192.168.10.0/24
PersistentKeepalive = 25
STUNMESH-go /tmp/config.yaml
---
refresh_interval: "1m"
log:
level: "debug"
interfaces:
"<utunX>":
peers:
"INTEL_MAC":
public_key: "<INTE_MAC_PUBLIC_KEY>"
plugin: cloudflare_main
stun:
address: "stun.l.google.com:19302"
plugins:
cloudflare_main:
type: cloudflare
zone_name: "<ZONE_NAME>"
api_token: "<API_TOKEN>"
Ping each other to check or you can use wg
to show the info.
You could use OSPF on Wireguard interface to create full mesh site-to-site VPN with dynamic routing.
Never be bothered to setup static route.
Wireguard interface didn't have link status (link up, down)
OSPF will say hello to remote peer periodically to check peer status.
It will also check wireguard's link status is up or not.
You can also reduce hello and dead interval in OSPF to make rapid response
Please also make sure setup access list or route map in OSPF to prevent redistribute public ip to remote peer.
It might cause to get incorrect route to remote peer endpoint and fail connect remote peer if you have multi-node.
BGP will only update when route table is changed.
It will take longer time to determine link status.
Not suggest to use BFD with BGP when router is small scale.
It will take too much overhead for link status detection
If you used this with your public network, and it's possible to enable VRF, please enable VRF with Wireguard interface.
Once you need Wireguard interface or private network to access internet.
Try to use VRF leaking to setup another default route to internet
Wireguard in Edgerouter
wireguard wg03 {
address <some route peer to peer IP>/30
description "to lab"
ip {
ospf {
network point-to-point
}
}
listen-port <wg port>
mtu 1420
peer <Remote Peer Public Key> {
allowed-ips 0.0.0.0/0
allowed-ips 224.0.0.5/32
persistent-keepalive 15
}
private-key ****************
route-allowed-ips false
}
OSPF in Edgerouter
policy {
access-list 1 {
description OSPF
rule 1 {
action permit
source {
inverse-mask 0.0.0.255
network <Your LAN CIDR>
}
}
rule 99 {
action deny
source {
any
}
}
}
}
protocols {
ospf {
access-list 1 {
export connected
}
area 0.0.0.0 {
network <Your network CIDR>
}
parameters {
abr-type cisco
router-id <Router ID>
}
passive-interface default
passive-interface-exclude <Your WG interface>
redistribute {
connected {
metric-type 2
}
}
}
}
- one shot command
- auto execute when routing engine notify change
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.