nhrpd is an implementation of the :abbr:NHRP (Next Hop Routing Protocol). NHRP is described in :rfc`2332`.
NHRP is used to improve the efficiency of routing computer network traffic over :abbr:`NBMA (Non-Broadcast, Multiple Access)` networks. NHRP provides an ARP-like solution that allows a system to dynamically learn the NBMA address of the other systems that are part of that network, allowing these systems to directly communicate without requiring traffic to use an intermediate hop.
Cisco Dynamic Multipoint VPN (DMVPN) is based on NHRP, and |PACKAGE_NAME| nhrpd implements this scenario.
nhrpd never handles routing of prefixes itself. You need to run some real routing protocol (e.g. BGP) to advertise routes over the tunnels. What nhrpd does it establishes 'shortcut routes' that optimizes the routing protocol to avoid going through extra nodes in NBMA GRE mesh.
nhrpd does route NHRP domain addresses individually using per-host prefixes. This is similar to Cisco FlexVPN; but in contrast to opennhrp which uses a generic subnet route.
To create NBMA GRE tunnel you might use the following (Linux terminal commands)::
ip tunnel add gre1 mode gre key 42 ttl 64 ip addr add 10.255.255.2/32 dev gre1 ip link set gre1 up
Note that the IP-address is assigned as host prefix to gre1. nhrpd will automatically create additional host routes pointing to gre1 when a connection with these hosts is established.
The gre1 subnet prefix should be announced by routing protocol from the hub nodes (e.g. BGP 'network' announce). This allows the routing protocol to decide which is the closest hub and determine the relay hub on prefix basis when direct tunnel is not established.
nhrpd will redistribute directly connected neighbors to zebra. Within hub nodes, these routes should be internally redistributed using some routing protocol (e.g. iBGP) to allow hubs to be able to relay all traffic.
This can be achieved in hubs with the following bgp configuration (network command defines the GRE subnet):
router bgp 65555 address-family ipv4 unicast network 172.16.0.0/16 redistribute nhrp exit-address-family
In addition to routing nhrp redistributed host prefixes, the hub nodes are also responsible to send NHRP Traffic Indication messages that trigger creation of the shortcut tunnels.
nhrpd sends Traffic Indication messages based on network traffic captured using NFLOG. Typically you want to send Traffic Indications for network traffic that is routed from gre1 back to gre1 in rate limited manner. This can be achieved with the following iptables rule.
iptables -A FORWARD -i gre1 -o gre1 \\ -m hashlimit --hashlimit-upto 4/minute --hashlimit-burst 1 \\ --hashlimit-mode srcip,dstip --hashlimit-srcmask 24 --hashlimit-dstmask 24 \\ --hashlimit-name loglimit-0 -j NFLOG --nflog-group 1 --nflog-range 128
You can fine tune the src/dstmask according to the prefix lengths you announce internal, add additional IP range matches, or rate limitation if needed. However, the above should be good in most cases.
This kernel NFLOG target's nflog-group is configured in global nhrp config with:
nhrp nflog-group 1
To start sending these traffic notices out from hubs, use the nhrp per-interface directive:
interface gre1 ip nhrp redirect
Integration with IKE
nhrpd needs tight integration with IKE daemon for various reasons. Currently only strongSwan is supported as IKE daemon.
nhrpd connects to strongSwan using VICI protocol based on UNIX socket (hardcoded now as /var/run/charon.vici).
strongSwan currently needs few patches applied. Please check out the http://git.alpinelinux.org/cgit/user/tteras/strongswan/log/?h=tteras-release,release and http://git.alpinelinux.org/cgit/user/tteras/strongswan/log/?h=tteras,working tree git repositories for the patches.