CentOS 7, As A 6in4 Server

Home » CentOS » CentOS 7, As A 6in4 Server
CentOS 3 Comments

Hello,

I have a VPS at a hoster where I got 3 /64 ipv6 prefixes/subnets, that are routed;

one for the VPS itself  – let us call this  srvprefix one for the tunnel, only ::1 (server side) and ::2 (home side) are used
– let us call this tunnelprefix and one for my network at home – let us call this homeprefix

now I’m just in test state, a CentOS VM is the other end of the tunnel;
(when the server runs well, my CentOS ZBOX will become the other end of the tunnel)

at the server

the eth0 device has  serverprefix::1, the sit1 device has tunnelprefix::1

the routing is set with /etc/sysconfig/network-scripts/route6-sit1

tunnelprefix::2 dev sit1
homeprefix::/64 via tunnelprefix::2 dev sit1

in sysctl.conf these are set

net.ipv6.conf.all.forwarding = 1
net.ipv6.conf.all.proxy_ndp = 1

now I have to do these

ip -6 neigh add proxy homeprefix::1 dev eth0
ip -6 neigh add proxy homeprefix::### dev eth0

the question, can I do something to avoid these “ip -6 neigh …”? if yes, what? and how?
can the hoster do something? if yes, what?

Thanks,

Walter

my ISP told me that he won’t deploy IPv6 within the next 5 years;

3 thoughts on - CentOS 7, As A 6in4 Server

  • I may be missing something, but you have 3 different networks, shouldn’t you just configure routing instead of using proxy_ndp?

    Regards.


    Roberto Ragusa mail at robertoragusa.it

  • can you specify this?
    yes, my own network at home, the network of the tunnel, and public the network where the VPS is part of;

    without these the following   is not possible,   -> Destination host unreachable

    ping6  homeprefix::1
    ping6 tunnelprefix::2
    ping6 tunnelprefix::1   (the sit1 device of the server itself)

    Thanks,

    Walter

  • If I understand correctly your hoster has assigned you 3 subnetworks but it is not routing them to your VPS, it just expects your VPS to use them on its interface. In this case, yes, you need proxy_ndp, and if there is a way to avoid enumerating each IP I would be interested to know.

    If instead tunnelprefix and homeprefix were just externally routed to serverprefix::1
    everything would be simpler, but that is not under your control, I assume.

    Regards.


    Roberto Ragusa mail at robertoragusa.it