Eric de Thouars Tom Eastep 2003-2004 2008 Eric de Thouars and Tom Eastep Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, with no Front-Cover, and with no Back-Cover Texts. A copy of the license is included in the section entitled GNU Free Documentation License. 6to4 tunneling with Shorewall can be used to connect your IPv6 network to another IPv6 network over an IPv4 infrastructure. It can also allow you to experiment with IPv6 even if your ISP doesn't provide IPv6 connectivity. More information on Linux and IPv6 can be found in the Linux IPv6 HOWTO. Details on how to setup a 6to4 tunnels are described in the section Setup of 6to4 tunnels.

6to4 tunnels provide a good way to introduce yourself to IPv6. Shorewall6 was developed on a network whose only IPv6 connectivity was an 6to4 Tunnel; that network is described in the remainder of this section. What is shown here requires Shorewall6 4.2.4 or later.

I have created an init script to make the job of configuring your firewall for IPv6 easier. The script is installed in /etc/init.d and configures ipv6, including a 6to4 tunnel, at boot time. The script works on OpenSuSE 11.0 and may need modification for other distributions. On OpenSuSE, the script is installed by copying it to /etc/init.d/ then running the command 'chkconfig --add ipv6'. At the top of the script, you will see several variables: SIT="sit1" INTERFACES="eth1" ADDRESS4=206.124.146.180 SLA=1 SIT - The name of the tunnel device. Usually 'sit1' INTERFACES - local interfaces that you want to configure for IPv6 ADDRESS4 - A static IPv4 address on your firewall that you want to use for the tunnel. SLA - The identity of the first local sub-network that you want to assign to the interfaces listed in INTERFACES. Normally one (0001). Here is the file from my firewall: SIT="sit1" INTERFACES="eth0 eth2" ADDRESS4=206.124.146.180 SLA=1 eth0 is the interface to my local network (both wired and wireless). eth2 goes to my DMZ which holds a single server. Here is a diagram of the IPv4 network: Here is the configuration after IPv6 is configured; the part in bold font is configured by the /etc/init.d/ipv6 script. gateway:~ # ip -6 addr ls 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 2002:ce7c:92b4:1::1/64 scope global valid_lft forever preferred_lft forever inet6 fe80::202:e3ff:fe08:55fa/64 scope link valid_lft forever preferred_lft forever 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 fe80::202:e3ff:fe08:484c/64 scope link valid_lft forever preferred_lft forever 4: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 2002:ce7c:92b4:2::1/64 scope global valid_lft forever preferred_lft forever inet6 fe80::2a0:ccff:fed2:353a/64 scope link valid_lft forever preferred_lft forever 24: sit1@NONE: <NOARP,UP,LOWER_UP> mtu 1480 inet6 ::206.124.146.180/128 scope global valid_lft forever preferred_lft forever inet6 2002:ce7c:92b4::1/128 scope global valid_lft forever preferred_lft forever gateway:~ # ip -6 route ls ::/96 via :: dev sit1 metric 256 expires 21333315sec mtu 1480 advmss 1420 hoplimit 4294967295 2002:ce7c:92b4::1 dev sit1 metric 256 expires 21333315sec mtu 1480 advmss 1420 hoplimit 4294967295 2002:ce7c:92b4:1::/64 dev eth0 metric 256 expires 21333315sec mtu 1500 advmss 1440 hoplimit 4294967295 2002:ce7c:92b4:2::/64 dev eth2 metric 256 expires 21333315sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev eth0 metric 256 expires 20748424sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev eth1 metric 256 expires 20748431sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev eth2 metric 256 expires 20748431sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev sit1 metric 256 expires 21333315sec mtu 1480 advmss 1420 hoplimit 4294967295 default via ::192.88.99.1 dev sit1 metric 1 expires 21333315sec mtu 1480 advmss 1420 hoplimit 4294967295 gateway:~ # You will notice that sit1, eth0 and eth2 each have an IPv6 address beginning with 2002: -- All 6to4 IPv6 addresses have that in their most significant 16 bits. The next 32-bits (ce7c:92b4) encode the IPv4 ADDRESS (206.124.146.180). So once you start the 6to4 tunnel, you are the proud owner of 280 IPv6 addresses! In the case shown here, 2002:ce7c:92b4::/48. The SLA is used to assign each interface in INTERFACES, a subnet of 264 addresses; in the case of eth0, 2002:ce7c:92b4:1::/64. I run radvd on the firewall to allow hosts conntected to eth0 and eth2 to automatically perform their own IPv6 configuration. Here is my /etc/radvd.conf file: interface eth0 { AdvSendAdvert on; MinRtrAdvInterval 3; MaxRtrAdvInterval 10; prefix 2002:ce7c:92b4:1::/64 { AdvOnLink on; AdvAutonomous on; AdvRouterAddr off; }; RDNSS 2002:ce7c:92b4:2:2a0:ccff:fedb:31c4 { AdvRDNSSOpen on; AdvRDNSSPreference 2; }; }; interface eth2 { AdvSendAdvert on; MinRtrAdvInterval 3; MaxRtrAdvInterval 10; prefix 2002:ce7c:92b4:2::/64 { AdvOnLink on; AdvAutonomous on; AdvRouterAddr off; }; RDNSS 2002:ce7c:92b4:2:2a0:ccff:fedb:31c4 { AdvRDNSSOpen on; AdvRDNSSPreference 2; }; }; Here is the automatic IPv6 configuration on my server attached to eth2: webadmin@lists:~/ftpsite/contrib/IPv6> /sbin/ip -6 addr ls 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 2002:ce7c:92b4:2:2a0:ccff:fedb:31c4/64 scope global dynamic valid_lft 2591995sec preferred_lft 604795sec inet6 fe80::2a0:ccff:fedb:31c4/64 scope link valid_lft forever preferred_lft forever webadmin@lists:~/ftpsite/contrib/IPv6> /sbin/ip -6 route ls 2002:ce7c:92b4:2::/64 dev eth2 proto kernel metric 256 expires 2592161sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev eth2 metric 256 expires 20746963sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 dev ifb0 metric 256 expires 20746985sec mtu 1500 advmss 1440 hoplimit 4294967295 default via fe80::2a0:ccff:fed2:353a dev eth2 proto kernel metric 1024 expires 29sec mtu 1500 advmss 1440 hoplimit 64 webadmin@lists:~/ftpsite/contrib/IPv6> You will note that the public IPv6 address of eth2 (2002:ce7c:92b4:2:2a0:ccff:fedb:31c4) was formed by concatenating the prefix for eth2 shown in radvd.conf (2002:ce7c:92b4:2) and the lower 64 bits of the link level address of eth2 (2a0:ccff:fedb:31c4). You will also notice that the address 2002:ce7c:92b4:2:2a0:ccff:fedb:31c4 appears in the RDNSS clauses in radvd.conf; that causes my server to be automatically configured as a DNS server. The default route is described using the link level address of eth2 on the firewall (fe80::2a0:ccff:fed2:353a). On my laptop, ursa: ursa:~ # ip -6 addr ls dev wlan0 3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 2002:ce7c:92b4:1:21a:73ff:fedb:8c35/64 scope global dynamic valid_lft 2591996sec preferred_lft 604796sec inet6 fe80::21a:73ff:fedb:8c35/64 scope link valid_lft forever preferred_lft forever ursa:~ # ip -6 addr ls dev eth0 3: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qlen 1000 inet6 2002:ce7c:92b4:1:21a:24ff:fecb:2bcc/64 scope global dynamic valid_lft 2591996sec preferred_lft 604796sec inet6 fe80::21a:73ff:fedb:8c35/64 scope link valid_lft forever preferred_lft forever ursa:~ # ip -6 route ls dev wlan0 2002:ce7c:92b4:1::/64 proto kernel metric 256 expires 2592160sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 metric 256 expires 21314573sec mtu 1500 advmss 1440 hoplimit 4294967295 default via fe80::202:e3ff:fe08:55fa proto kernel metric 1024 expires 28sec mtu 1500 advmss 1440 hoplimit 64 ursa:~ # ip -6 route ls dev eth0 2002:ce7c:92b4:1::/64 proto kernel metric 256 expires 2592160sec mtu 1500 advmss 1440 hoplimit 4294967295 fe80::/64 metric 256 expires 21314573sec mtu 1500 advmss 1440 hoplimit 4294967295 default via fe80::202:e3ff:fe08:55fa proto kernel metric 1024 expires 28sec mtu 1500 advmss 1440 hoplimit 64 ursa:~ # Here is the resulting simple IPv6 Network:

We need to add an entry in /etc/shorewall/tunnels and restart Shorewall: #TYPE ZONE GATEWAY GATEWAY # ZONE 6to4 net #LAST LINE -- ADD YOUR ENTRIES BEFORE THIS ONE -- DO NOT REMOVE

STOP -- If you have followed the instructions above, you should have a completely functional IPv6 network. Try: ping6 2001:19f0:feee::dead:beef:cafe If that doesn't work from your firewall and from any local IPv6 systems that you have behind your firewall, do not go any further until it does work. If you ask for help from the Shorewall team, the first question we will ask is 'With Shorewall6 cleared, can you ping6 2001:19f0:feee::dead:beef:cafe?'. The Shorewall6 configuration on my firewall is a very basic three-interface one. Key entry in /etc/shorewall6/shorewall6.conf: IP_FORWARDING=On /etc/shorewall6/zones: #ZONE TYPE OPTIONS IN OUT # OPTIONS OPTIONS fw firewall net ipv6 loc ipv6 dmz ipv6 #LAST LINE -- ADD YOUR ENTRIES BEFORE THIS ONE -- DO NOT REMOVE /etc/shorewall6/interfaces: #ZONE INTERFACE BROADCAST OPTIONS net sit1 detect tcpflags,forward=1,nosmurfs loc eth0 detect tcpflags,forward=1 dmz eth2 detect tcpflags,forward=1 #LAST LINE -- ADD YOUR ENTRIES BEFORE THIS ONE -- DO NOT REMOVE /etc/shorewall6/policy: #SOURCE DEST POLICY LOG LEVEL LIMIT:BURST net all DROP info loc net ACCEPT dmz net ACCEPT all all REJECT info /etc/shorewall6/rules: #ACTION SOURCE DEST PROTO DEST SOURCE ORIGINAL RATE USER/ MARK # PORT PORT(S) DEST LIMIT GROUP # # Accept DNS connections from the firewall to the network # DNS/ACCEPT $FW net # # Accept SSH connections from the local network for administration # SSH/ACCEPT loc $FW # # Allow Ping everywhere # Ping/ACCEPT all all # #LAST LINE -- ADD YOUR ENTRIES BEFORE THIS ONE -- DO NOT REMOVE

Suppose that we have the following situation: We want systems in the 2002:100:333::/64 subnetwork to be able to communicate with the systems in the 2002:488:999::/64 network. This is accomplished through use of the /etc/shorewall/tunnels file and the ip utility for network interface and routing configuration. Unlike GRE and IPIP tunneling, the /etc/shorewall/policy, /etc/shorewall/interfaces and /etc/shorewall/zones files are not used. There is no need to declare a zone to represent the remote IPv6 network. This remote network is not visible on IPv4 interfaces and to iptables. All that is visible on the IPv4 level is an IPv4 stream which contains IPv6 traffic. Separate IPv6 interfaces and ip6tables rules need to be defined to handle this traffic. In /etc/shorewall/tunnels on system A, we need the following: #TYPE ZONE GATEWAY GATEWAY ZONE 6to4 net 134.28.54.2 This entry in /etc/shorewall/tunnels opens the firewall so that the IPv6 encapsulation protocol (41) will be accepted to/from the remote gateway. Use the following commands to setup system A: >ip tunnel add tun6to4 mode sit ttl 254 remote 134.28.54.2 >ip link set dev tun6to4 up >ip addr add 3ffe:8280:0:2001::1/64 dev tun6to4 >ip route add 2002:488:999::/64 via 3ffe:8280:0:2001::2 Similarly, in /etc/shorewall/tunnels on system B we have: #TYPE ZONE GATEWAY GATEWAY ZONE 6to4 net 206.191.148.9 And use the following commands to setup system B: >ip tunnel add tun6to4 mode sit ttl 254 remote 206.191.148.9 >ip link set dev tun6to4 up >ip addr add 3ffe:8280:0:2001::2/64 dev tun6to4 >ip route add 2002:100:333::/64 via 3ffe:8280:0:2001::1 On both systems, restart Shorewall and issue the configuration commands as listed above. The systems in both IPv6 subnetworks can now talk to each other using IPv6.