Q. How do I configure static IPv6 networking under RHEL 5.x / Fedora / CentOS Linux?
A. Red Hat / CentOS / Fedora RHEL support IPv6 out of box. All you have to do is update two files and turn on networking.
You need to update and configure following files for IPv6 configuration:
- /etc/sysconfig/network : Turn on networking in this file.
- /etc/sysconfig/network-scripts/ifcfg-eth0 : Set default IPv6 router IP and server IP address in this file.
Open /etc/sysconfig/network file, enter:
# vi /etc/sysconfig/network
Append following line:
NETWORKING_IPV6=yes |
Open /etc/sysconfig/network-scripts/ifcfg-eth0 (1st network config file)
# vi /etc/sysconfig/network-scripts/ifcfg-eth0
Append following config directives for IPv6:
IPV6INIT=yes IPV6ADDR=<IPv6-IP-Address> IPV6_DEFAULTGW=<IPv6-IP-Gateway-Address> |
Here is my sample file with mix of IPv4 and IPv6 assigned to eth0:
DEVICE=eth0 BOOTPROTO=static ONBOOT=yes HWADDR=00:30:48:33:bc:33 IPADDR=202.54.1.5 GATEWAY=202.54.1.3 NETMASK=255.255.255.248 IPV6INIT=yes IPV6ADDR=2607:f0d0:1002:0011:0000:0000:0000:0002 IPV6_DEFAULTGW=2607:f0d0:1002:0011:0000:0000:0000:0001 |
Where,
- NETWORKING_IPV6=yes|no – Enable or disable global IPv6 initialization.
- IPV6INIT=yes – Enable or disable IPv6 configuration for all interfaces.
- IPV6ADDR=2607:f0d0:1002:0011:0000:0000:0000:0002 – Specify a primary static IPv6 address here.
- IPV6_DEFAULTGW=2607:f0d0:1002:0011:0000:0000:0000:0001 – Add a default route through specified gateway.
Save and close the file. Restart networking:
# service network restart
Verify your configuration by pinging ipv6 enabled site such as ipv6.google.com:
$ ping6 ipv6.google.com
Sample output:
PING ipv6.google.com(2001:4860:b002::68) 56 data bytes 64 bytes from 2001:4860:b002::68: icmp_seq=1 ttl=59 time=93.2 ms 64 bytes from 2001:4860:b002::68: icmp_seq=2 ttl=59 time=95.0 ms 64 bytes from 2001:4860:b002::68: icmp_seq=3 ttl=59 time=94.2 ms 64 bytes from 2001:4860:b002::68: icmp_seq=4 ttl=59 time=95.2 ms 64 bytes from 2001:4860:b002::68: icmp_seq=5 ttl=59 time=94.8 ms 64 bytes from 2001:4860:b002::68: icmp_seq=6 ttl=59 time=95.1 ms 64 bytes from 2001:4860:b002::68: icmp_seq=7 ttl=59 time=93.3 ms 64 bytes from 2001:4860:b002::68: icmp_seq=8 ttl=59 time=93.8 ms --- ipv6.google.com ping statistics --- 8 packets transmitted, 8 received, 0% packet loss, time 7010ms rtt min/avg/max/mdev = 93.268/94.376/95.268/0.799 ms
Traces path to a network host, enter:
$ traceroute6 ipv6.google.com
Print default IPv6 routing table, enter:
$ route -n -A inet6
Sample output:
Kernel IPv6 routing table Destination Next Hop Flags Metric Ref Use Iface ::1/128 :: U 0 42531 1 lo ::62.41.14.144/128 :: U 0 0 1 lo ::127.0.0.1/128 :: U 0 0 1 lo ::/96 :: U 256 0 0 sit0 2001:470:1f04:55a::2/128 :: U 0 15201 1 lo 2001:470:1f04:55a::/64 :: U 256 0 0 sit1 fe80::4833:22f4/128 :: U 0 0 1 lo fe80::212:3fff:fe75:fa0d/128 :: U 0 0 1 lo fe80::/64 :: U 256 0 0 eth0 fe80::/64 :: U 256 0 0 sit1 ff00::/8 :: U 256 0 0 eth0 ff00::/8 :: U 256 0 0 sit1 ::/0 :: U 1 0 0 sit1
Once IPv6 configured properly, you need to setup IPv6 firewall using ip6tables command under Linux.
you have no ipv6 mask :(
Thanks so much for this article. It is exactly what I have been looking for. I will recommend this article. Keep up the good work.
Nice reminders, but I’d like to make two suggestions:
1. Since the gateways are not specific to an interface, I keep the default gateway stuff (v4 and v6) in /etc/sysconfig/network.
2. I also don’t like the v6 autoconf for servers, so I turn it off in /etc/sysconfig/network: IPV6_AUTOCONF=no
Thanks for the quick tips.
Pete
So how would one find out there IPV6 address to use as a static one and what the default gw IPV6 address us to use?
I know the IPV4 addresses > how do i find its IPV6?
Thanks in advance
I try to add another interface (ifcfg-eth0:1) with the same IPv6 lines as in ifcfg-eth0, but the IPv6 does not appear when I use ifconfig. How can I assign an IPv6 to another network interface?
Use IPV6ADDR_SECONDARIES in ifcfg-eth0. Generally, v6 devices expect to have multiple addresses. If you do an ifconfig on eth0, you’ll probably see at least two v6 addresses already (one a link-local scope and one a global scope).
IPV6ADDR_SECONDARIES=”[/] …” (optional)
A list of secondary IPv6 addresses (e.g. useful for virtual hosting)
Example:
IPV6ADDR_SECONDARIES=”3ffe:ffff:0:1::10 3ffe:ffff:0:2::11/128″
Is there any directive to make network service start at boot time.
Try
I think that works for me, thanks Mr Vivek.
Just a question. How can I calculate usable hosts address? I have assigned IPv6 /48 network and need single adress for each server, for example
3x IPv6 address for server-1
2x IPv6 address for server-2
etc …
Check out a tool called sipcalc. It supports IPv4 and IPv6 and tells you all the relevant information about a netblock.
http://www.routemeister.net/projects/sipcalc/
Thanks Louis
ipv6.google.com doesn’t resolve today, so it’s a bad example. Try http://www.v6.facebook.com.
any idea how to configure IPv6 in bridge for RH based OS?
I have a question. If one has setup a CentOS server, and it is now in production sitting online in use 24/7 with IPv4, is it possible to add IPv6 to it without having to restart the network service?
What I have tried:
$ modprobe ipv6
$ ifconfig eth0 && ifconfig xenbr0
By this point there are no ipv6 addresses shown in ifconfig
$ ip addr add 2001:xx:xx:xx:1::2/64 dev xenbr0
By this point I still have no ipv6 addresses in ifconfig, all I get is:
RTNETLINK answers: Operation not supported
If I absolutely have to reboot the server then I will find a good point to do so and do it but if it’s even remotely possible to do without then that would be ideal. The server hosts Xen virtual machines using bridged networking configuration and somehow I doubt that restarting the network service would play nice with them and come back automatically at Virtual Machine level.
Hi I dont find /etc/sysconfig/network-scripts/ifcfg-eth0, I only have /etc/sysconfig/network-scripts/ifcfg-lo…….what can I do ? Help me please
George – how about cp /etc/sysconfig/network-scripts/ifcfg-lo /etc/sysconfig/network-scripts/ifcfg-eth0
?
The scripts aren’t really smart. If you have to use a link local address for your gateway, its not going to work because the route command doesn’t specify the interface.
I solved this quite bluntly – no clue on how future proof this is – by using a zone identifier on the gateway address. As in:
IPV6_DEFAULTGW=fe80::1%eth2
Before more people start the discussion, many hosting providers use the link local address as the gateway because of redundancy. In case of failure of one of the routers the link local address is transferred the global addresses these routers have are NOT! And hence, using one the global addresses at these providers means your routing redundancy is gone.
Earlier I commented on default IPv6 route not working with a link local address for the gateway and that it did work by adding the zone identifier (fe80::1%eth2 for example). Whilst this works nicely for me, I just found out it’s not the official way…
Please update your howto with the following ifcfg-ethX option:
IPV6_DEFAULTDEV=
From the docs:
IPV6_DEFAULTDEV= (optional)
Add a default route through specified interface without specifying next hop
Type of interface will be tested whether this is allowed
Examples:
IPV6_DEFAULTDEV=”eth0″ INVALID example!
IPV6_DEFAULTDEV=”ppp0″
IPV6_DEFAULTDEV=”sit1″
Examples for 6to4
IPV6_DEFAULTDEV=”tun6to4″
Add default route through dedicated 6to4 tunnel device “tun6to4”, if configured
As with all link local address it always has to be specified what device to use. Even ping6 has this issue, because all interfaces get an address in the same range (link local that is).
Hi,
I’m getting:
[ipv6_add_route] ‘No route to host’ adding route ‘::/0’ via gateway ‘2607:fcd0:ff03::1’ through device ”
When I restart network. Any ideas? I followed everything above!
HI,
I need to configuer ipv6 dhcp server with 112 perfix.Can You help me out
I prefer to NOT put in a default gateway for IPv6. You should let the routers on the subnet announce themselves. That’s what Router Announcements (part of Neighbor Discovery) are for. They will announce the proper IPv6 address (which should be a link-local fe80 address) as needed. That RA also includes the “prefix”, which is the IPv6 term for the netmask. Since IPv6 doesn’t permit non-contiguous subnet bits, a straight prefix, which will almost always be /64, should appear in the RA.
Nice