P2P-NAT-SOLUTION
- Status: raw
- Category: networking
- Editor: Antonio Antonino <antonio@status.im>
- Contributors:
- Álvaro Castro-Castilla <alvaro@status.im>
- Daniel Sanchez-Quiros <danielsq@status.im>
- Petar Radovic <petar@status.im>
- Gusto Bacvinka <augustinas@status.im>
- Youngjoon Lee <youngjoon@status.im>
- Filip Dimitrijevic <filip@status.im>
Abstract
This specification defines a comprehensive NAT (Network Address Translation) traversal solution for the Nomos P2P network. The solution enables nodes to automatically determine their NAT status and establish both outbound and inbound connections regardless of network configuration. The strategy combines AutoNAT, dynamic port mapping protocols, and continuous verification to maximize public reachability while maintaining decentralized operation.
Motivation
Network Address Translation presents a critical challenge for Nomos participants, particularly those operating on consumer hardware without technical expertise. The Nomos network requires a NAT traversal solution that:
- Automatic Operation: Works out-of-the-box without user configuration
- Inclusive Participation: Enables nodes on consumer hardware to participate effectively
- Decentralized Approach: Leverages the existing Nomos P2P network rather than centralized services
- Progressive Fallback: Escalates through increasingly complex protocols as needed
- Dynamic Adaptation: Handles changing network environments and configurations
The solution must ensure that nodes can both establish outbound connections and accept inbound connections from other peers, maintaining network connectivity across diverse NAT configurations.
Specification
Terminology
- Public Node: A node that is publicly reachable via a public IP address or valid port mapping
- Private Node: A node that is not publicly reachable due to NAT/firewall restrictions
- Dialing: The process of establishing a connection using the libp2p protocol stack
- NAT Status: Whether a node is publicly reachable or hidden behind NAT
Key Design Principles
Optional Configuration
The NAT traversal strategy must work out-of-the-box whenever possible. Users who do not want to engage in configuration should only need to install the node software package. However, users requiring full control must be able to configure every aspect of the strategy.
Decentralized Operation
The solution leverages the existing Nomos P2P network for coordination rather than relying on centralized third-party services. This maintains the decentralized nature of the network while providing necessary NAT traversal capabilities.
Progressive Fallback
The protocol begins with lightweight checks and escalates through more complex and resource-intensive protocols. Failure at any step moves the protocol to the next stage in the strategy, ensuring maximum compatibility across network configurations.
Dynamic Network Environment
Unless explicitly configured for static addresses, each node's public or private status is assumed to be dynamic. A once publicly-reachable node can become unreachable and vice versa, requiring continuous monitoring and adaptation.
Node Discovery Considerations
The Nomos public network encourages participation from a large number of nodes, many deployed through simple installation procedures. Some nodes will not achieve Public status, but the discovery protocol must track these peers and allow other nodes to discover them. This prevents network partitioning and ensures Private nodes remain accessible to other participants.
NAT Traversal Protocol
Protocol Requirements
Each node MUST:
- Run an AutoNAT client, except for nodes statically configured as Public
- Use the Identify protocol to advertise support for:
/nomos/autonat/2/dial-request
for main network/nomos-testnet/autonat/2/dial-request
for public testnet/nomos/autonat/2/dial-back
and/nomos-testnet/autonat/2/dial-back
respectively
NAT State Machine
The NAT traversal process follows a multi-phase state machine:
Phase Implementation
Phase 0: Bootstrapping and Identifying Public Nodes
If the node is statically configured by the operator to be Public, the procedure stops here.
The node utilizes bootstrapping and discovery mechanisms to find other Public nodes. The Identify protocol confirms which detected Public nodes support AutoNAT v2.
Phase 1: NAT Detection
The node starts an AutoNAT client and inspects its own addresses. For each public IP address, the node verifies public reachability via AutoNAT. If any public IP addresses are confirmed, the node assumes Public status and moves to Phase 3. Otherwise, it continues to Phase 2.
Phase 2: Automated Port Mapping
The node attempts to secure port mapping on the default gateway using:
- PCP (Port Control Protocol) - Most reliable
- NAT-PMP (NAT Port Mapping Protocol) - Second most reliable
- UPnP-IGD (Universal Plug and Play Internet Gateway Device) - Most widely deployed
Port Mapping Algorithm:
def try_port_mapping():
# Step 1: Get the local IPv4 address
local_ip = get_local_ipv4_address()
# Step 2: Get the default gateway IPv4 address
gateway_ip = get_default_gateway_address()
# Step 3: Abort if local or gateway IP could not be determined
if not local_ip or not gateway_ip:
return "Mapping failed: Unable to get local or gateway IPv4"
# Step 4: Probe the gateway for protocol support
supports_pcp = probe_pcp(gateway_ip)
supports_nat_pmp = probe_nat_pmp(gateway_ip)
supports_upnp = probe_upnp(gateway_ip) # Optional for logging
# Step 5-9: Try protocols in order of reliability
# PCP (most reliable) -> NAT-PMP -> UPnP -> fallback attempts
protocols = [
(supports_pcp, try_pcp_mapping),
(supports_nat_pmp, try_nat_pmp_mapping),
(True, try_upnp_mapping), # Always try UPnP
(not supports_pcp, try_pcp_mapping), # Fallback
(not supports_nat_pmp, try_nat_pmp_mapping) # Last resort
]
for supported, mapping_func in protocols:
if supported:
mapping = mapping_func(local_ip, gateway_ip)
if mapping:
return mapping
return "Mapping failed: No protocol succeeded"
If mapping succeeds, the node uses AutoNAT to confirm public reachability. Upon confirmation, the node assumes Public status. Otherwise, it assumes Private status.
Port Mapping Sequence:
Phase 3: Network Monitoring
Unless explicitly configured, nodes must monitor their network status and restart from Phase 1 when changes are detected.
Public Node Monitoring:
A Public node must restart when:
- AutoNAT client no longer confirms public reachability
- A previously successful port mapping is lost or refresh fails
Private Node Monitoring:
A Private node must restart when:
- It gains a new public IP address
- Port mapping is likely to succeed (gateway change, sufficient time passed)
Network Monitoring Sequence:
Public Node Responsibilities
A Public node MUST:
Run an AutoNAT server
Listen on and advertise via Identify protocol its publicly reachable multiaddresses:
/{public_peer_ip}/udp/{port}/quic-v1/p2p/{public_peer_id}
Periodically renew port mappings according to protocol recommendations
Maintain high availability for AutoNAT services
Peer Dialing
Other peers can always dial a Public peer using its publicly reachable multiaddresses:
/{public_peer_ip}/udp/{port}/quic-v1/p2p/{public_peer_id}
Implementation Requirements
Mandatory Components
All Nomos nodes MUST implement:
- AutoNAT client for NAT status detection
- Port mapping clients for PCP, NAT-PMP, and UPnP-IGD
- Identify protocol for capability advertisement
- Network monitoring for status change detection
Optional Enhancements
Nodes MAY implement:
- Custom port mapping retry strategies
- Enhanced network change detection
- Advanced AutoNAT server load balancing
- Backup connectivity mechanisms
Configuration Parameters
AutoNAT Configuration
autonat:
client:
dial_timeout: 15s
max_peer_addresses: 16
throttle_global_limit: 30
throttle_peer_limit: 3
server:
dial_timeout: 30s
max_peer_addresses: 16
throttle_global_limit: 30
throttle_peer_limit: 3
Port Mapping Configuration
port_mapping:
pcp:
timeout: 30s
lifetime: 7200s # 2 hours
retry_interval: 300s
nat_pmp:
timeout: 30s
lifetime: 7200s
retry_interval: 300s
upnp:
timeout: 30s
lease_duration: 7200s
retry_interval: 300s
Security Considerations
NAT Traversal Security
- Port Mapping Validation: Verify that requested port mappings are actually created
- AutoNAT Server Trust: Implement peer reputation for AutoNAT servers
- Gateway Communication: Secure communication with NAT devices
- Address Validation: Validate public addresses before advertisement
Privacy Considerations
- IP Address Exposure: Public nodes necessarily expose IP addresses
- Traffic Analysis: Monitor for patterns that could reveal node behavior
- Gateway Information: Minimize exposure of internal network topology
Denial of Service Protection
- AutoNAT Rate Limiting: Implement request throttling for AutoNAT services
- Port Mapping Abuse: Prevent excessive port mapping requests
- Resource Exhaustion: Limit concurrent NAT traversal attempts
Performance Characteristics
Scalability
- AutoNAT Server Load: Distributed across Public nodes
- Port Mapping Overhead: Minimal ongoing resource usage
- Network Monitoring: Efficient periodic checks
Reliability
- Fallback Mechanisms: Multiple protocols ensure high success rates
- Continuous Monitoring: Automatic recovery from connectivity loss
- Protocol Redundancy: Multiple port mapping protocols increase reliability
References
- Multiaddress spec
- Identify protocol spec
- AutoNAT v2 protocol spec
- Circuit Relay v2 protocol spec
- PCP - RFC 6887
- NAT-PMP - RFC 6886
- UPnP IGD - RFC 6970
Copyright
Copyright and related rights waived via CC0.