Since Hetzner pretty much doesn't give out any IPv4 addresses, no matter what the reason is, I decided to give IPv6 a try.

These is the assigned IP address subnet mask

IPs: 2a01:4f8:190:5112:: /64
Gateway: fe80::1
Useable IP addresses:
2a01:4f8:190:5112::2 to 2a01:4f8:190:5112:ffff:ffff:ffff:ffff

First I added IP with

ip -6 addr add 2a01:4f8:190:5112::2/64 dev eth0

added this block to /etc/network/interfaces

iface eth0 inet6 static
  address 2a01:4f8:190:5112::2
  netmask 64
  gateway fe80::1

and restarted networking.

When I attempt to ping this IP from the server, it pings ok.

root@server ~ # ping6 2a01:4f8:190:5112::2
PING 2a01:4f8:190:5112::2(2a01:4f8:190:5112::2) 56 data bytes
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=1 ttl=64 time=0.035 ms
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=2 ttl=64 time=0.040 ms
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=3 ttl=64 time=0.041 ms

however any attempts to do pinging from outside the server, result in

connect: Network is unreachable

I am new to configuring IPv6 addresses so it's possible I may have forgotten something. I am running Ubuntu 13.10 if it matters.

Output from ip6tables -L -n -v

Chain INPUT (policy ACCEPT 10383 packets, 2639K bytes)
 pkts bytes target     prot opt in     out     source               destination         

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
 pkts bytes target     prot opt in     out     source               destination         

Chain OUTPUT (policy ACCEPT 13967 packets, 1119K bytes)
 pkts bytes target     prot opt in     out     source               destination         

Thank you very much in advance, any help is much appreciated.

  • Can you paste the output from ip6tables -L -n -v into your question?
    – MadHatter
    May 2, 2014 at 9:15
  • Sure. Updated my question.
    – Screatch
    May 2, 2014 at 9:16
  • Ah, OK, that's not it, then!
    – MadHatter
    May 2, 2014 at 9:17
  • What about pinging out from the server? ping6 -n www.free.fr.
    – MadHatter
    May 2, 2014 at 9:18
  • 2
    Does the computer where you are pinging from have IPv6? May 2, 2014 at 9:25

2 Answers 2


Nothing is wrong. This just in, from my v6-enabled server:

[me@lory mail]$ ping6 2a01:4f8:190:5112::2
PING 2a01:4f8:190:5112::2(2a01:4f8:190:5112::2) 56 data bytes
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=1 ttl=53 time=21.2 ms
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=2 ttl=53 time=21.1 ms
64 bytes from 2a01:4f8:190:5112::2: icmp_seq=3 ttl=53 time=21.1 ms

I suspect the problem may be ipv6 on the machine you're PINGing from, rather than the one you're PINGing to. But in any case, that's not what you asked; to answer the question you asked, your server's ipv6 is set up just fine.

Though if I were you, I'd put up some kind of v6-firewall ruleset fairly quickly!

Edit: Is it supported? Well, that sort of depends on whether you've set it up! v6 works perfectly well, but here in the UK, despite being fairly standard in data centres it's not yet common amongst domestic ISPs. There are some who can do it (I use one), but you often have to ask for it. Even if you have one that supports it, and you ask for it and get it, your gateway device will need to be told about it, and will have to run some kind of route advertisement daemon and either hand out addresses or support stateless v6 address autoconfiguration so that LAN clients get the right addresses and routes.

So: yes, it can be supported, but both your and your ISP have to be involved. It doesn't somehow magically all work out of the box, with free sparkle ponies as an added bonus.

By the way, may I thank you for not redacting your ip addresses in this question? It was a very great deal easier to answer as a result of your putting in the real addresses.

  • This makes sense. I just tried to do pinging to this address from 3 different servers and 4 separate PC's (3 ubuntu and 1 windows), none of them were able to ping IP address, is IPv6 support really this poor at the moment?
    – Screatch
    May 2, 2014 at 9:28
  • 2
    @Screatch Yes, it really is that poor. According to Google, IPv6 deployment on clients is at about 3% (google.com/intl/en/ipv6/statistics.html). According to my calculations we needed to reach that point about 13 years ago in order to complete the upgrade to IPv6 on time (kasperd.net/~kasperd/transition). It is almost certainly your ISP, you have to blame for not having IPv6 on your client machines.
    – kasperd
    May 2, 2014 at 9:35
  • 2
    It's an excellent time to go dual-stack, and we all should. It's too early to go single-stack v6, though, except where you are servicing a known v6-enabled client population.
    – MadHatter
    May 2, 2014 at 9:53
  • 1
    That was kasperd's point. What exactly is your business need for a second v4 address?
    – MadHatter
    May 2, 2014 at 11:18
  • 1
    APNIC (Asia-Pacific) and RIPE NCC (Europe, the Middle East and parts of Central Asia) have run out of unallocated IPv4 addresses a long time ago (2012). ISPs cannot get any new addresses (unless they find someone who want to sell their addresses, but prices will be high), so they will be very careful with giving addresses to customers. ARIN (North America) is going to run out soon as well. Countries like Romania, Switzerland and Belgium have many residential IPv6 users, but other countries are far far behind. So we are in a situation where we are out of IPv4, but cannot go IPv6-only yet :( May 3, 2014 at 21:22

In order to use IPv6, you need to have it not only on the server, but also on the client. If your ISP is not willing to do their job and upgrade their network, you can look into using a 6in4 tunnel. There are different providers, some do well in some areas, others do well in other areas. Overall my feeling is that Hurricane Electric is the best choice.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .