Our network has been based around a home-built router for quite some time, ever since I got fed up with the crappy ActionTec router that Verizon bundled with our FiOS service. (If you’re going to offer high-speed internet, you should probably bundle equipment that can actually keep up.) I had originally followed a slightly older version of these instructions to get a nice basic router going. But I finally wanted better. I wanted the bright, shiny, new thing. I wanted IPv6.
So, here’s my instructions for going from an existing IPv4 router to dual-stack IPv4/6.
Note: I am using dnsmasq for DNS and DHCP, hostapd for wireless management, and an iptables firewall. Since Verizon still doesn’t widely support consumer IPv6, I’m using a tunnel broker to get my /6 address. If you’re using a different setup your mileage may vary. If you find anything that I appear to have forgotten, please let me know!
Step 1: Recompile the Kernel
This should be obvious: if you want to run ipv6 you need ipv6 support in your kernel. In order to trim as much off my kernel as possible I did not have it built in, and had to recompile.
You should also add netfilter support for ipv6 so that your firewall will work.
Networking support --->
Networking options --->
<*> The IPv6 protocol --->
<*> IPv6: IPv6-in-IPv4 tunnel (SIT driver)
[*] Network packet filtering framework (Netfilter) --->
IPv6: Netfilter Configuration --->
<M> IPv6 NAT
<M> IP6 tables support (required for filtering)
<M> Packet filtering
<M> ip6tables NAT support
<M> MASQUERADE target support
... other filtering options as you may need for your situation
Step 2: Update Your IPv6 Support
Again, it was never compiled in, in order to trim off unused bits of code. Add ‘ipv6’ to your USE variable and emerge --newuse world
Step 3: Install network tools (if they aren’t already)
emerge --noreplace sys-apps/iproute2 net-firewall/iptables
Step 4 (optional): Set up your tunnel
If your ISP doesn’t provide ipv6, and many don’t, you need to request an address range from a tunnel broker. I’m using Hurricane Electric, which is free, but there are others — see this list or just google it.
If you have multiple machines on your network (which is assumed, since this is a router guide), you may prefer a /48, so that autoconfig works nicely, instead of the default /64. This guide assumes a /48.
Going forward, replace 2001:470:891a: with your own /48 range.
Now activate your tunnel:
ip tunnel add he-ipv6 mode sit remote 1.2.3.4 local 5.6.7.8 ttl 255 ip link set he-ipv6 up ip addr add 2001:470:1f06:2a3::2/64 dev he-ipv6 ip route add ::/0 dev he-ipv6 ip -f inet6 addr
Step 5: Update Your Net Config
I have two wired and one wireless card in my router. Here’s what my /etc/conf.d/net looks like:
# enp2s0 is my exterior wired nic (aka public facing) # enp3s5 is my interior wired nic # wlp3s6 is my interior wireless nic dhcp_enp2s0="nodns" # we choose our own DNS, tyvm config_enp3s5="192.168.0.1/24 2001:470:891a:0::/64" modules_wlp3s6="!iwconfig !wpa_supplicant" config_wlp3s6="192.168.1.1/24 2001:470:891a:1::/48" dns_servers_wlp3s6="127.0.0.1"
After making appropriate changes, restart your NICs. If you’re working remotely, you may want to be connected via two paths instead of just one (so when you inevitably get bounced and can’t reconnect, you still have a way back in).
A properly-configured set of addresses looks like this:
# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: enp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 00:e0:4d:bf:03:f5 brd ff:ff:ff:ff:ff:ff
inet 108.20.118.17/24 brd 108.20.118.255 scope global enp2s0
valid_lft forever preferred_lft forever
inet6 fe80::cbdf:25c0:c948:f4bb/64 scope link
valid_lft forever preferred_lft forever
3: enp3s5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UNKNOWN group default qlen 1000
link/ether 00:04:5a:42:a6:98 brd ff:ff:ff:ff:ff:ff
inet 192.168.0.1/24 brd 192.168.0.255 scope global enp3s5
valid_lft forever preferred_lft forever
inet6 2001:470:891a::/64 scope global
valid_lft forever preferred_lft forever
inet6 fe80::204:5aff:fe42:a698/64 scope link
valid_lft forever preferred_lft forever
4: wlp3s6: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
link/ether 00:1a:ef:07:4d:a7 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.1/24 brd 192.168.1.255 scope global wlp3s6
valid_lft forever preferred_lft forever
inet6 2001:470:891a:1::/64 scope global
valid_lft forever preferred_lft forever
inet6 fe80::21a:efff:fe07:4da7/64 scope link
valid_lft forever preferred_lft forever
5: sit0@NONE: mtu 1480 qdisc noop state DOWN group default
link/sit 0.0.0.0 brd 0.0.0.0
6: he-ipv6@NONE: <POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1480 qdisc noqueue state UNKNOWN group default
link/sit 108.20.118.17 peer 209.51.161.14
inet6 2001:470:1f06:2a3::2/64 scope global
valid_lft forever preferred_lft forever
inet6 fe80::6c14:7611/64 scope link
valid_lft forever preferred_lft forever
Test it with a ping:
ping6 www.kame.net
Step 6: Reconfigure dnsmasq
You’ll need to add router advertisments and your new addresses. Rather than hard-coding an address, dnsmasq offers a ‘constructor’ label which figures it out automatically. Here’s the relevant section from my /etc/dnsmasq.conf:
domain-needed bogus-priv domain=jonesling.us dhcp-authoritative enable-ra dhcp-range=192.168.0.20,192.168.0.100,12h dhcp-range=192.168.1.20,192.168.1.100,12h dhcp-range=192.168.2.20,192.168.2.100,12h dhcp-range=::,constructor:enp3s5,ra-names,slaac,12h dhcp-range=::,constructor:wlp3s6,ra-names,slaac,12h resolv-file=/etc/resolv.dnsmasq selfmx enable-ra
And restart it: /etc/init.d/dnsmasq restart
Step 7: Configure your firewall
Pretty much every iptables reference in your firewall config will be mirrored with an ip6tables command.
Here’s my script to set up iptables (if you see an error or something stupid, I would appreciate your criticism – paired with your reasoning on why it should be changed so I can know better for next time).
#!/bin/bash
# based on http://www.gentoo.org/doc/en/home-router-howto.xml
# set to '0' to lock the kids out
OPEN_INTERNET=1
# these systems can get shut out when OPEN_INTERNET isn't true
declare -a NO_SURFING=( 'wii-u'
'kids-computer'
)
# these systems never get shut out
declare -a OK_SURFING=( 'parents-computer'
'parents-phone'
)
# these ports take precedence over CLOSED_PORTS
declare -a OPEN_TCP_PORTS=( 'ssh'
'http'
'mail'
'submission'
)
declare -a OPEN_UDP_PORTS=( 'submission' )
# if the port is meant to be closed, we close tcp *AND* udp
declare -a CLOSED_PORTS=( '0:1055'
'svn'
'distcc'
'x11'
'nfs'
'icpv2'
'mysql'
'rtsp'
'3128' # squid
'3130' # squid ICP
'3551' # nisport
)
declare -a LAN_SERVICES=( "svn" )
# blacklisted IPs and ranges
# http://www.iana.org/assignments/ipv4-address-space/ipv4-address-space.xhtml
declare -a IP_BLACKLIST=( # APINIC
# AFRINIC
# LACNIC
...
)
LAN=enp3s5
WLAN=wlp3s6
WAN=enp2s0
SIT=he-ipv6
INSIDE=( $LAN $WLAN )
LOCAL_RANGE_IPV4='192.168.0.0/16'
LOCAL_RANGE_IPV6='2001:470:891a::'
# First we flush our current rules
iptables -F
iptables -t nat -F
ip6tables -F
ip6tables -t nat -F
# Setup default policies to handle unmatched traffic
iptables -P INPUT ACCEPT
iptables -P OUTPUT ACCEPT
iptables -P FORWARD DROP
ip6tables -P INPUT ACCEPT
ip6tables -P OUTPUT ACCEPT
ip6tables -P FORWARD DROP
# Then we lock our services so they only work from the LAN
iptables -I INPUT 1 -i ${LAN} -j ACCEPT
iptables -I INPUT 1 -i ${WLAN} -j ACCEPT
iptables -I INPUT 1 -i lo -j ACCEPT
ip6tables -I INPUT 1 -i ${LAN} -j ACCEPT
ip6tables -I INPUT 1 -i ${WLAN} -j ACCEPT
ip6tables -I INPUT 1 -i lo -j ACCEPT
# block members of IP_BLACKLIST, plus any addresses passed in on the
# command line
for IP in ${IP_BLACKLIST[@]} ; do
iptables -I INPUT -s ${IP} -p TCP --dport ssh -j DROP
done
for IP in $@; do
iptables -I INPUT -s ${IP} -d 0/0 -j REJECT
done
iptables -A INPUT -p UDP --dport bootps -i ${WAN} -j REJECT
ip6tables -A INPUT -p UDP --dport bootps -i ${SIT} -j REJECT
iptables -A INPUT -p UDP --dport domain -i ${WAN} -j REJECT
ip6tables -A INPUT -p UDP --dport domain -i ${SIT} -j REJECT
# Explicitly allow access to services on the WAN
for SERVICE in ${LAN_SERVICES[@]} ; do
for IFACE in ${INSIDE[@]} ; do
iptables -A INPUT -p TCP --dport svn -i ${IFACE} -j ACCEPT
iptables -A INPUT -p UDP --dport svn -i ${IFACE} -j ACCEPT
ip6tables -A INPUT -p TCP --dport svn -i ${IFACE} -j ACCEPT
ip6tables -A INPUT -p UDP --dport svn -i ${IFACE} -j ACCEPT
done
done
# Allow access to our server from the WAN
for PORT in ${OPEN_TCP_PORTS[@]} ; do
iptables -A INPUT -p TCP --dport $PORT -i ${WAN} -j ACCEPT
ip6tables -A INPUT -p TCP --dport $PORT -i ${SIT} -j ACCEPT
done
for PORT in ${OPEN_UPD_PORTS[@]} ; do
iptables -A INPUT -p UDP --dport PORT -i ${WAN} -j ACCEPT
ip6tables -A INPUT -p UDP --dport PORT -i ${SIT} -j ACCEPT
done
# Drop TCP / UDP packets to privileged ports
for PORT in ${CLOSED_PORTS[@]} ; do
iptables -A INPUT -p TCP -i ${WAN} -d 0/0 --dport ${PORT} -j DROP
ip6tables -A INPUT -p TCP -i ${SIT} -d 0/0 --dport ${PORT} -j DROP
iptables -A INPUT -p UDP -i ${WAN} -d 0/0 --dport ${PORT} -j DROP
ip6tables -A INPUT -p UDP -i ${SIT} -d 0/0 --dport ${PORT} -j DROP
done
iptables -I FORWARD -i ${LAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
iptables -A FORWARD -i ${LAN} -s $LOCAL_RANGE_IPV4 -j ACCEPT
ip6tables -I FORWARD -i ${LAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
ip6tables -A FORWARD -i ${LAN} -s $LOCAL_RANGE_IPV6 -j ACCEPT
if (( OPEN_INTERNET )); then
echo 'yay, everybody gets internet'
iptables -I FORWARD -i ${WLAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
iptables -A FORWARD -i ${WLAN} -s $LOCAL_RANGE_IPV4 -j ACCEPT
ip6tables -I FORWARD -i ${WLAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
ip6tables -A FORWARD -i ${WLAN} -s $LOCAL_RANGE_IPV6 -j ACCEPT
else
echo "boo, only ${OK_SURFING[@]} get internet"
for IP in ${OK_SURFING[@]}; do
iptables -I FORWARD -i ${WLAN} -d $IP/255.255.255.255 -j ACCEPT
iptables -A FORWARD -i ${WLAN} -s $IP/255.255.255.255 -j ACCEPT
done
fi
iptables -A FORWARD -i ${WAN} -d $LOCAL_RANGE_IPV4 -j ACCEPT
ip6tables -A FORWARD -i ${WAN} -d $LOCAL_RANGE_IPV6 -j ACCEPT
iptables -t nat -A POSTROUTING -o ${WAN} -j MASQUERADE
ip6tables -t nat -A POSTROUTING -o ${SIT} -j MASQUERADE
# This is so when we boot we don't have to run the rules by hand
/etc/init.d/iptables save
/etc/init.d/ip6tables save
# fail2ban should be reloaded after flushing iptables
/etc/init.d/fail2ban reload
Step 8: Update your DNS
Add a AAAA record to your domain’s DNS record. You may have to keep this one up-to-date yourself.
Interesting to note: you might be thinking “crap, what’s the ipv6 equivalent of these CNAME records?” Stop worrying, there isn’t. The CNAME is read like normal, but ipv6 clients will then look up the AAAA (instead of the A) record of the destination host. It just works.
What? You built your own router but you don’t have your own domain? WTF is wrong with you?
Step 9: Reboot your clients
While I was working, I made a bunch of mistakes and my clients had multiple ipv6 addresses – making networking from them unstable as they didn’t necessarily know which address to use. Rebooting will clear them – and make sure your config is proper.
At this point your clients should be in ipv6 and you’re gonna be all excited to see if work. Browsers take ipv6 addresses a little differently: http://[2001:470:1f06:2a3::2]/
