Håvards page

Building a Debian GNU/Linux IPv6 home router

Feb 9, 2013

Introduction

This short post describes how i configured my own IPv6 home router using Debian GNU/Linux. I used a Dreamplug, but any form of device with at least 2 NICs should be usable. Allthough this guide describes the setup using Debian, it should be no problem using another distribution or one of the BSD variants.

I will not go into to many details about the configurations, for more information you should read the man pages and/or documentation.

Sources

This blog post from Phil Dibowitz was very helpful setting me on the right track and gives you more details about what is going on. Also I recommend reading Recommended Simple Security Capabilities in Customer Premises Equipment for Providing Residential IPv6 Internet Service.

DNS

At the minimum you want to have a recursive server for your clients. I also configured a zone and allowing dynamic updates so that dhcpd can add clients and also adding the routers own IP-addresses.

Installing bind

aptitude install bind9

Creating a key for dynamic updates

ddns-confgen -a hmac-md5 -r test -r /dev/urandom -q -k dhcp_updater > /etc/bind/dns-dhcp.key
chown root:bind /etc/bind/dns-dhcp.key
chown 640 /etc/bind/dns-dhcp.key

/etc/bind/named.conf.options

options {
        directory "/var/cache/bind";

        forwarders { };

        //========================================================================
        // If BIND logs error messages about the root key being expired,
        // you will need to update your keys.  See https://www.isc.org/bind-keys
        //========================================================================
        dnssec-validation yes;

        auth-nxdomain no;    # conform to RFC1035
        listen-on-v6 { any; };

        max-cache-size 44040192;
};

The empty forwarders statement will be filled in later by a post-up script. Set max-cache-size to a suitable size or leave it out to use the defaults.

Setting up zones

I’m using bifrost.haavard.name as my DNS zone for the network and 192.168.0.0/24 for the internal IPv4 network. My router is called heimdal. Change accordingly for your needs or skip this if you don’t want an internal DNS zone.

mkdir /etc/bind/zones
chown root:bind /etc/bind/zones
chmod 2770 /etc/bind/zones
cat<<"EOD" > /etc/bind/zones/db.bifrost.haavard.name
$ORIGIN .
$TTL 3600       ; 1 hour
bifrost.haavard.name    IN SOA  heimdal.bifrost.haavard.name. post.haavard.name. (
                                1025       ; serial
                                604800     ; refresh (1 week)
                                86400      ; retry (1 day)
                                2419200    ; expire (4 weeks)
                                86400      ; minimum (1 day)
                                )
                        NS      heimdal.bifrost.haavard.name.
heimdal                 A       192.168.0.1
EOD
cat<<"EOD" > /etc/bind/zones/db.192.168.0.0
$ORIGIN .
$TTL 3600       ; 1 hour
0.168.192.in-addr.arpa  IN SOA  heimdal.bifrost.haavard.name. post.haavard.name. (
                                6          ; serial
                                604800     ; refresh (1 week)
                                86400      ; retry (1 day)
                                2419200    ; expire (4 weeks)
                                86400      ; minimum (1 day)
                                )
                        NS      heimdal.bifrost.haavard.name.
EOD
chown bind:bind /etc/bind/zones/*

/etc/bind/named.conf.local

include "/etc/bind/zones.rfc1918";

include "/etc/bind/bind.keys";

include "/etc/bind/dns-dhcp.key";

zone "bifrost.haavard.name" {
  type master;
  file "/etc/bind/zones/db.bifrost.haavard.name";
  allow-update { key dhcp_updater; };
};

zone "0.168.192.in-addr.arpa" {
  type master;
  file "/etc/bind/zones/db.192.168.0";
  allow-update { key dhcp_updater; };
};

/usr/local/bin/update-bind-forwarders

This script is used in the post up section of ifup to update the forwarders for bind.

#!/bin/bash

forwarders=$(egrep -v '127.0.0.1|::1' /etc/resolv.conf | awk '/^nameserver/ {print $2}' | tr '\n' ';')
sed -ri "s/forwarders[[:space:]]+\{[^}]+\};/forwarders { $forwarders };/" /etc/bind/named.conf.options
/etc/init.d/bind9 reload

DHCP server for IPv4

Installing dhcp server

aptitude install isc-dhcp-server

/etc/dhcpd.conf

ddns-update-style interim;
ddns-domainname "bifrost.haavard.name";

include "/etc/bind/dns-dhcp.key";

zone bifrost.haavard.name. {
   primary 127.0.0.1;
   key dhcp_updater;
}

zone 0.168.192.in-addr.arpa. {
   primary 127.0.0.1;
   key dhcp_updater;
}

option domain-name "bifrost.haavard.name haavard.name";
option domain-name-servers 192.168.0.1;

default-lease-time 86400;
max-lease-time 720000;

authoritative;

log-facility local7;

subnet 192.168.0.0 netmask 255.255.255.0 {
  range 192.168.0.50 192.168.0.198;
  option routers 192.168.0.1;
}

If you want to give some hosts static addresses, for instance for port forwarding, add to the config file

host XX {
  hardware ethernet XX:XX:XX:XX:XX:XX;
  fixed-address 192.168.0.X;
}

Router config

We’re going to run a post-script in dhcp which will configure addresses on the network interfaces and update DNS. To avoid hard coding this script, I use a config file which I put in /etc/router.conf. If you do not want to use a local DNS zone, skip the dns_* configuration parameters. NB! The dns_search parameter is required if you want to use RDNSS.

/etc/router.conf

# the internet facing interface
external_nic = eth0
# space seperated list of internal nics
internal_nics = eth1

ipv4_prefix = 192.168.0.0/24

dns_update_server = ::1
dns_update_key = /etc/bind/dns-dhcp.key
dns_ttl = 3600
dns_zone = bifrost.haavard.name
dns_external_name = heimdal-ext
dns_eth1_name = heimdal
dns_search = bifrost.haavard.name haavard.name

radvd

I have choosen to use stateless autoconfig for IPv6. This is handled by the radvd daemon.

Installing radvd

aptitude install radvd

/etc/radvd.conf.tmpl

This is the template used by the dhcp post script to configure radvd. Modify as needed. If you do not want to use RDNSS remove the RDNSS and DNSSL lines.

interface __IFACE__ {
   AdvSendAdvert on;
   prefix __PREFIX__
   {
     AdvOnLink on;
     AdvAutonomous on;
     AdvRouterAddr on;
   };

   RDNSS __IP__ { };
   DNSSL __SEARCH__ {};
};

Multicast routing daemon

You probably want to run a multicast routing daemon, I’ve choosen to use mrd6.

Installing mrd6

aptitude install mrd6

DHCP post script

This script will assign IP-addresses to the internal nics from the assigned address range, configure radvd as well as populate dns

The scripts in /etc/dhcp/dhclient-enter-hooks.d are not executables, but shell scripts being sourced by dhclient-scripts. That is why we need to create the actual script which is written in perl somewhere else and just call it from here.

Required packages

aptitude install libconfig-file-perl libnet-ip-perl

/etc/dhcp/dhclient-enter-hooks.d/router

/usr/local/bin/dhclient-router

/usr/local/bin/dhclient-router

#!/usr/bin/perl

use strict;
use Config::File;
use File::Temp;
use Net::IP;

my $config = Config::File::read_config_file('/etc/router.conf');

exit 0 unless $ENV{'interface'} = $config->{'external_nic'};
exit 0 unless $ENV{'reason'} =~ /^(BOUND|REBIND|RENEW)6?$/;

if(exists $ENV{'new_ip_address'} or exists $ENV{'new_ip6_address'}) {
  my $recordtype;
  my $ip;
  if(exists $ENV{'new_ip_address'}) {
    $recordtype = 'A';
    $ip = $ENV{'new_ip_address'};
  } else {
    $recordtype = 'AAAA';
    $ip = $ENV{'new_ip6_address'};
  };
  if(exists $config->{'dns_update_server'}
      and exists $config->{'dns_update_key'}
      and exists $config->{'dns_ttl'}
      and exists $config->{'dns_zone'}
      and exists $config->{'dns_external_name'}) {
    open(my $nsupdate, "| nsupdate -k $config->{'dns_update_key'}");
    print $nsupdate "server $config->{'dns_update_server'}\n";
    print $nsupdate "zone $config->{'dns_zone'}\n";
    print $nsupdate "update delete $config->{'dns_external_name'}.$config->{'dns_zone'}. IN $recordtype\n";
    print $nsupdate "update add $config->{'dns_external_name'}.$config->{'dns_zone'}. $config->{'dns_ttl'} IN $recordtype $ip\n";
    print $nsupdate "send\n";
    print $nsupdate "quit\n";
    close($nsupdate);
  };
};

exit 0 unless exists $ENV{'new_ip6_prefix'};

my $tempconf = File::Temp->new;
$tempconf->unlink_on_destroy(0);
open(my $template, "/etc/radvd.conf.tmpl") or die "Unable to open /etc/radvd.conf.tmpl: $!";
my @template = <$template>;
close($template);

my @internal_nics = split(/\s+/, $config->{'internal_nics'});
my $num_internal_nets = scalar(@internal_nics);

my $prefix = Net::IP->new($ENV{'new_ip6_prefix'});
my @octets = split(/:/, $prefix->ip);
my $start_octet = hex($octets[3]);
my $last_octet = hex((split(/:/, $prefix->last_ip))[3]);

die "Not enough addresses for all internal nics" unless(($last_octet - $start_octet) >= $num_internal_nets);

if(exists $config->{'dns_update_server'}
    and exists $config->{'dns_update_key'}
    and exists $config->{'dns_ttl'}
    and exists $config->{'dns_zone'}) {
  open(my $nsupdate, "| nsupdate -k $config->{'dns_update_key'}");
  print $nsupdate "server $config->{'dns_update_server'}\n";
  print $nsupdate "zone $config->{'dns_zone'}\n";
  print $nsupdate "update delete $config->{'dns_zone'}. IN APL\n";
  my $apl_entry;
  if(exists $config->{'ipv4_prefix'}) {
    $apl_entry = "1:$config->{'ipv4_prefix'} 2:$ENV{'new_ip6_prefix'}";
  } else {
    $apl_entry = "2:$ENV{'new_ip6_prefix'}";
  };
  print $nsupdate "update add $config->{'dns_zone'}. $config->{'dns_ttl'} IN APL $apl_entry\n";
  print $nsupdate "send\n";
  print $nsupdate "quit\n";
  close($nsupdate);
};

my $i = 0;
foreach my $internal_nic (@internal_nics) {
  my $new_ip = Net::IP->new(sprintf("%s:%s:%s:%x::1", $octets[0], $octets[1], $octets[2], $start_octet + $i));
  my $new_prefix = sprintf("%s:%s:%s:%x::/64", $octets[0], $octets[1], $octets[2], $start_octet + $i++);
  my $current_ip = undef;
  open(my $addr, "ip -6 addr show dev $internal_nic scope global |");
  while(<$addr>) {
    if(/inet6 ([0-9a-f:]+)\/64 scope global\s*$/) {
      $current_ip = Net::IP->new($1);
    };
  };

  my $ip = $new_ip->ip;
  my $search = $config->{'dns_search'};
  foreach my $line (@template) {
    my $l = $line;
    $l =~ s/__IFACE__/$internal_nic/;
    $l =~ s/__PREFIX__/$new_prefix/;
    $l =~ s/__IP__/$ip/;
    $l =~ s/__SEARCH__/$search/;
    print $tempconf $l;
  };

  if(defined $current_ip) {
    next if $new_ip->overlaps($current_ip) == $IP_IDENTICAL; # address already set
    my $ip = $current_ip->ip;
    `ip -6 addr del $ip/64 dev $internal_nic`;
  };
  my $ip = $new_ip->ip;
  `ip -6 addr add $ip/64 dev $internal_nic`;

  if(exists $config->{'dns_update_server'}
      and exists $config->{'dns_update_key'}
      and exists $config->{'dns_zone'}
      and exists $config->{'dns_ttl'}
      and exists $config->{'dns_' . $internal_nic . '_name'}) {
    open(my $nsupdate, "| nsupdate -k $config->{'dns_update_key'}");
    print $nsupdate "server $config->{'dns_update_server'}\n";
    print $nsupdate "zone $config->{'dns_zone'}\n";
    my $entry = $config->{'dns_' . $internal_nic . '_name'} . '.' . $config->{'dns_zone'} . '.';
    print $nsupdate "update delete $entry IN AAAA\n";
    print $nsupdate "update add $entry $config->{'dns_ttl'} IN AAAA $ip\n";
    print $nsupdate "send\n";
    print $nsupdate "quit\n";
    close($nsupdate);
  };
};

my $tempconf_filename = $tempconf->filename;
close($tempconf);
unless(system("diff /etc/radvd.conf $tempconf_filename >/dev/null") == 0) {
  rename($tempconf_filename, '/etc/radvd.conf');
  chmod 0644, $tempconf_filename;
  `/etc/init.d/radvd restart >/dev/null` ;
} else {
  unlink($tempconf_filename);
};

Firewall

Especially for IPv6 there are som important firewall rules as discussed in Recommended Simple Security Capabilities in Customer Premises Equipment for Providing Residential IPv6 Internet Service. and for IPv4 you have to setup NAT. The rules here should address these issues. I have added some commented out examples for port forwarding and openings for specific ports, you will have to tailor the firewall configuration to suit your own needs. The IPv4 configuration is for running the 192.168.0.0/24 network on eth1. The iptables rules are for one external nic on eth0 and internal on eth1, you have to modify to fit your topology.

/etc/iptables

*nat
:PREROUTING ACCEPT [0:0]
:INPUT ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
:POSTROUTING ACCEPT [2:120]
# Port forward port 22 to 192.168.0.199. Also remember to rule in the FORWARD table
#-A PREROUTING -i eth0 -p tcp -m tcp --dport 22 -j DNAT --to-destination 192.168.0.199
-A POSTROUTING -o eth0 -j MASQUERADE
COMMIT
*filter
:INPUT DROP [0:0]
:FORWARD DROP [0:0]
:OUTPUT ACCEPT [0:0]
:bad_tcp_packets - [0:0]
-A bad_tcp_packets -p tcp --tcp-flags SYN,ACK SYN,ACK -m conntrack --ctstate NEW -j REJECT --reject-with tcp-reset
-A bad_tcp_packets -p tcp ! --syn -m conntrack --ctstate NEW -j DROP

-A INPUT -p tcp -j bad_tcp_packets
-A INPUT -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
# allow traffic from internal nic
-A INPUT -s 192.168.0.0/24 -i eth1 -j ACCEPT
# allow dhcp on external nic
-A INPUT -i eth0 -p udp -m udp --sport 67 --dport 68 -j ACCEPT
-A INPUT -i lo -j ACCEPT
-A INPUT -p icmp -j ACCEPT
# allow DNS and NTP
-A INPUT -p udp -m udp --sport 53 -j ACCEPT
-A INPUT -p udp -m udp --sport 123 -j ACCEPT
-A INPUT -m limit --limit 11/minute -j LOG --log-level info --log-prefix "IPv4-INPUT "
-A INPUT -j REJECT

-A FORWARD -p tcp -j bad_tcp_packets
# allow forwarding for internal nic
-A FORWARD -i eth1 -j ACCEPT
-A FORWARD -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
# Port forward port 22 to 192.168.0.199
#-A FORWARD -d 192.168.0.199/32 -p tcp --dport 22 -j ACCEPT
-A FORWARD -m limit --limit 11/minute -j LOG --log-level info --log-prefix "IPv4-FORWARD "
-A FORWARD -j REJECT --reject-with icmp-host-unreachable

-A OUTPUT -p tcp -j bad_tcp_packets
COMMIT

/etc/ip6tables

*filter
:INPUT DROP [0:0]
:FORWARD DROP [0:0]
:OUTPUT ACCEPT [27:2428]
:bad_tcp_packets - [0:0]
-A bad_tcp_packets -p tcp --tcp-flags SYN,ACK SYN,ACK -m conntrack --ctstate NEW -j REJECT --reject-with tcp-reset
#-A bad_tcp_packets -p tcp ! --syn -m conntrack --ctstate NEW -j LOG --log-prefix "New not syn:"
-A bad_tcp_packets -p tcp ! --syn -m conntrack --ctstate NEW -j DROP

-A INPUT -p tcp -j bad_tcp_packets
-A INPUT -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
# allow dhcp
-A INPUT -i eth0 -p udp -m udp --dport dhcpv6-client -j ACCEPT
# allow icmp, essential for IPv6!
-A INPUT -p icmpv6 -j ACCEPT
# allow traffic from internal nic
-A INPUT -i eth1 -j ACCEPT
-A INPUT -i lo -j ACCEPT
# allow link local multicast
-A INPUT --dest ff02::/16 -j ACCEPT
# allow DNS and NTP
-A INPUT -p udp -m udp --sport 53 -j ACCEPT
-A INPUT -p udp -m udp --sport 123 -j ACCEPT
-A INPUT -m limit --limit 11/minute -j LOG --log-level info --log-prefix "IPv6-INPUT "
-A INPUT -j REJECT

-A FORWARD -p tcp -j bad_tcp_packets
# multicast, allow global multicast prefix, reject all else
-A FORWARD --source ff00::/8 -j REJECT --reject-with icmp6-adm-prohibited
-A FORWARD --dest ff0e::/16 -j ACCEPT
-A FORWARD --dest ff1e::/16 -j ACCEPT
-A FORWARD --dest ff2e::/16 -j ACCEPT
-A FORWARD --dest ff3e::/16 -j ACCEPT
-A FORWARD --dest ff4e::/16 -j ACCEPT
-A FORWARD --dest ff5e::/16 -j ACCEPT
-A FORWARD --dest ff6e::/16 -j ACCEPT
-A FORWARD --dest ff7e::/16 -j ACCEPT
-A FORWARD --dest ff00::/8 -j REJECT --reject-with icmp6-no-route
# link-local
-A FORWARD --dest fe80::/10 -j DROP
-A FORWARD --source fe80::/10 -j DROP
-A FORWARD --dest 3fa::/10 -j REJECT --reject-with icmp6-no-route
-A FORWARD --source 3fa::/10 -j REJECT --reject-with icmp6-no-route
# site local
-A FORWARD --dest 3fb::/10 -j REJECT --reject-with icmp6-no-route
-A FORWARD --source 3fb::/10 -j REJECT --reject-with icmp6-no-route
# ipv4-mapped
-A FORWARD --dest ::FFFF:0:0/96 -j REJECT --reject-with icmp6-no-route
-A FORWARD --source ::FFFF:0:0/96 -j REJECT --reject-with icmp6-no-route
# documentation prefix
-A FORWARD --dest 2001:db8::/32 -j REJECT --reject-with icmp6-no-route
-A FORWARD --source 2001:db8::/32 -j REJECT --reject-with icmp6-no-route
# orchid
-A FORWARD --dest 2001:10::/28 -j REJECT --reject-with icmp6-no-route
-A FORWARD --source 2001:10::/28 -j REJECT --reject-with icmp6-no-route

# routing header 0, deprecated
-A FORWARD -m rt --rt-type 0  -j REJECT --reject-with icmp6-adm-prohibited

# allow forwarding for internal nic
-A FORWARD -i eth1 -j ACCEPT
-A FORWARD -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT

# icmp
-A FORWARD -p icmpv6 -j ACCEPT

# ipsec
-A FORWARD -p udp --dport 500 -j ACCEPT
-A FORWARD -m ipv6header --soft --header esp -j ACCEPT
-A FORWARD -m ipv6header --soft --header auth -j ACCEPT

# Example if you want to allow ssh to your internal machines
#-A FORWARD -p tcp --dport 22 -j ACCEPT

-A FORWARD -m limit --limit 11/minute -j LOG --log-level info --log-prefix "IPv6-FORWARD "
-A FORWARD -j REJECT --reject-with icmp6-addr-unreachable

-A OUTPUT -p tcp -j bad_tcp_packets
# link-local
-A OUTPUT --dest 3fa::/10 -j REJECT --reject-with icmp6-no-route
-A OUTPUT --source 3fa::/10 -j REJECT --reject-with icmp6-no-route
# site local
-A OUTPUT --dest 3fb::/10 -j REJECT --reject-with icmp6-no-route
-A OUTPUT --source 3fb::/10 -j REJECT --reject-with icmp6-no-route
# ipv4-mapped
-A OUTPUT --dest ::FFFF:0:0/96 -j REJECT --reject-with icmp6-no-route
-A OUTPUT --source ::FFFF:0:0/96 -j REJECT --reject-with icmp6-no-route
# documentation prefix
-A OUTPUT --dest 2001:db8::/32 -j REJECT --reject-with icmp6-no-route
-A OUTPUT --source 2001:db8::/32 -j REJECT --reject-with icmp6-no-route
# orchid
-A OUTPUT --dest 2001:10::/28 -j REJECT --reject-with icmp6-no-route
-A OUTPUT --source 2001:10::/28 -j REJECT --reject-with icmp6-no-route

COMMIT

/etc/network/if-pre-up.d/iptables

#!/bin/bash

/sbin/iptables-restore < /etc/iptables
/sbin/ip6tables-restore < /etc/ip6tables

sysctl config

/etc/sysctl.conf

net.ipv4.ip_forward=1

    net.ipv6.conf.all.forwarding=1

    net.ipv4.tcp_ecn = 1

Network config

We have to manually run dhclient using post-up as there is per now no built in ifup method in Debian to run dhclient with -P which is needed for prefix delegation.

/etc/network/interfaces

auto lo
iface lo inet loopback

auto eth0
iface eth0 inet dhcp
  pre-up /etc/init.d/bind9 start
iface eth0 inet6 auto
  pre-up /etc/init.d/bind9 start
  post-up /sbin/sysctl net.ipv6.conf.eth0.accept_ra=2
  post-up /sbin/dhclient -1 -6 -P -N -pf /run/dhclient.eth0-ipv6.pid -lf /var/lib/dhcp/dhclient.eth0-ipv6.leases eth0
  post-up /usr/local/bin/update-bind-forwarders
  pre-down /sbin/dhclient -v -r -pf /run/dhclient.eth0-ipv6.pid -lf /var/lib/dhcp/dhclient.eth0-ipv6.leases eth0

auto eth1
iface eth1 inet static
  address 192.168.0.1
  netmask 255.255.255.0
  pre-up /etc/init.d/bind9 start
iface eth1 inet6 manual

Using CoDel queue algorithm

To avoid bufferbloat you can use the CoDel active queue management algorithm. This requires Linux kernel 3.5 or higher as well as iproute 20121001-1 or newer.

Installation

Download debloat.sh from https://github.com/dtaht/deBloat and place it in /etc/network/if-up.d/debloat. Remember to install ethtool if it is not already installed.