Hello everyone.
I’ve recently decided that IPv6 is the future (a bit late I know)
I’ve implemented IPv6 on my cloud servers at OVH. We get a full /64 (that’s what is said in the doc. But if you follow the doc, then your gateway is out of the subnet. In fact, you get a /56 even tough they say /64)
Now, I’m trying to implement IPv6 in my LAN’s (multiple ones. In fact, I have 8 vlan’s on my home network. Quite useless I know but I like it)
Our ISP’s in Belgium are futureproof (at least on that point, I will not comment on the bandwidth/volume/pricing) and give IPv6 to everyone. In fact, we are one of the most developed country’s in IPv6 deployment if I read correctly.
So, our ISP is giving us /64 trough DHCPV6-PD. I ve putted some magic commands in my router to get the said /64 on my PPPOE interface (ISP side) and it worked. My router is pingable from the internet with V6 \o/
Problem is that even tough my router is getting an IPv6 IP and a /64 subnet, I have no idea on how to redistribute them. I entered some commands that I’ve read on multiple guides that were talking about SLAAC and now only get fe80 addresses in my LAN’s.
Has someone already done this? Am I missing something? do I even search in the right direction?
Every single help I can get is welcome 
I’m new with IPv6
Dhcp distributes addresses so dhcpv6 server needed
This link to setting up dhcp, replace ip4 with your ip6
How does my laptop get an fe80 address if there is no DHCPv6 already enabled (it gets none if nothing is configured on the router side)
There might be another problem because of the IPv6 subnet wich is dynamicaly re attributed. So I’m a little bit scared that making a DHCPv6 server might break the config…
But while I’m thinking at it, if my router can get a V6 address trough DHCP, shoul I try to make a DHCP relay?
Vlan for ip4 with dhcp
Vlan for ip6 with dhcp
Bridge
If netsh theres several ip6toip4 techniques not sure with edge os
A more indepth form is sorting ip4 to standard ethernet and ip6 run as fcoe but any path between you and server that poe connection and power is used need ip4toip6 for those hops or ip4 can be funky in its behaivor
But running 2 instances of dhcp on same eth port will be fine if (netsh my example) netsh add interface “lan” add subinterface ip4 add subinterface ip6
Setup each sub as if other didnt exist(dhcp dns firewall etc)
Note add wlan as subinterface under subs for ip# or build same as lan under int=wlan then make virtual interface and add sub=lan add sub wlan
I don’t know what happened but it started working at one point…
So, I changed stuff.
I tried making a dhcp relay, it didn’t help. After that, I tried adding a fe80 interface on one of the lan sides and it automaticaly got an ipv6 in the /64 range of the PPPOE interface both on my router interface and my laptop connected to the LAN.
Ping didn’t worked tough. So I deleted the DHCP relay. Still nothing. At that point, I was able to ping between fe80’s of the same LAN. After that, public IPv6 internet was available like magic and then it crashed again.
So to this point, I have a IPv6/64 on PPPOE and LAN that are in the same range while having fe80’s in each net. Laptop is receiving public IPv6 address but while the PPPOE ipv6 interface is pingable from the outside world and stable to ping, if I try to ping in the LAN, it works, then not, then again and is looping like that and I cannot find the reason why.
Just in case you wonder, my routes on the laptop are looking like this :
arnold@altea:~$ ip -6 route
::1 dev lo proto kernel metric 256 pref medium
2a02:a03f:3e00:a008::/64 dev wlp3s0 proto ra metric 600 pref medium
fe80::/64 dev wlp3s0 proto kernel metric 600 pref medium
default via fe80::c002:11de:f2a9:bac6 dev wlp3s0 proto ra metric 600 pref medium
fe80 address is pingable
Try metric 500’s
Lookup minimum segment size(mss)
Set mtu ip4=1484 ip6=1280 as its not changeable per standards
Mss ≥200 something ≤580?
Mss and mtu and metric are easy to confuse
Especially mtu and metric as some guides explain metric to be mtu while others make it seem its id # (im thinking vlan rules or vci vdi thing)
I leave windows metric under adapter set to auto but set under netsh int ip# set int (several submenus and options), mss, mtu, frag rules, receive buffer size(set in long bit/byte form so if 50mbps then 50mB10241024=byte*8=bit,each 1024 drops mega to kilo to base)
Note netsh has bug that will not hold receive buffer setting
Ill lookup the setting im describing and post in few mins
netsh » interface » ipv4 » set »
Global » reassemblylimit(not held through reboot no matter store=used )
Interface » # goodies
Subinterface » explained above
Theres ip6 too
Netsh » interface has 6to4 and ip4toip6 ip4toip4 ip6toip6 ip6toip4 settings as well
Find these under edge os linux? Equivs
Apparently, it’s not a MTU problem.
I think it must be a route problem.
Currently, I m having a public IPv6 interface on the PPPOE interface and LAN interface as well as a fe80 address.
I can ping the public IP address on the LAN interface from the LAN as well as the fe80 local interface. But I cannot ping the PPPOE public IPv6 interface from the LAN
Update :
I restarted the router.
I get a public IPv6 address on the PPPOE interface and on the laptop. I can ping the fe80 internaly with the laptop but not the public internet or even the router public IPv6 address.
What is strange tough is that my default gateway is correct. But even if I try to ping the fe80 of the PPPOE interface from the laptop, I get no response at all… It must be at the router level
Update :
a traceroute to a public ipv6 address get lost before the first node. As if the computer didn’t know what to do even tough ip -6 route command shows the correct LAN ipv6 gateway.
(It’s not due to the laptop, I get the same problem on others systems and tested my laptop 2 hours ago on a functionnal IPv6 setup, no trouble at all)
Are you using a docsis device that carries a ip4 dsl backbone while?
Check out hitron america and see cgn 2200 docsis all in one device pdf
Its 2 pages but lists techs that may be useful for you to look for in your device as i dont try to remember every accronym
ICMPv6 enabled on the device firewall? Generally not by default in windows.
*edited for query clarification.
ICMPv6 working on my device (I use arch btw 
)
It’s not Docsis, it’s VDSL2 but our ISP is kind of nazi with their protocols. In fact, there are 2 ISP’s in belgium and one is owned by the state so they basicaly do whatever they want. We don’t even have more info about how they are implementing IPv6. We just know that the hardware they give to us is pretty bad.
(I’m starting to wonder if it won’t be easyer to simply become my own ISP with BGP true a tunnel true my ISP’s network to get a static IPv6 pool that can be correctly implemented)
Okay. This is getting ridiculous.
Without touching to the config, the entire ipv6 public ip addresses aren’t provided anymore in the LAN.
I followed this guide in the first place (and adapted it to my network) https://medium.com/@nurblieh/ipv6-on-the-edgerouter-lite-c95e3cc8d49d
I still don’t know how to solve the problem with this IPv6 doing what it wants when it wants. Et even worked at one point for 5 minutes and I didn’t touched it since… (I did tough but everytime have putted it back to initial state)
Think big hotel
Cable line faster than phone
Docsis allows ip6 host to become a dslbackbone out the phone jack on device thus manage 100+rooms as dsl clients(means voip phone as well to room)
Your local phone company could be doing same thing
Docsis 1gbit ÷ 10mbit clients=# clients
If this is case then you need loopback for your private ip6 to be routed thru ip4 and recieved sameway, ip6toip4 for outbound, ip4toip6 for inbound
I don’t think that our ISP is doing that. We have VDSL2 to the home that goes directly to the street box. I know for a fact that there is real single mode fiber in the street box for almost every single street. It’s relatively stable and well done.
Just that the hardware they provide directly to the end client is kind of bad (source code leaked once so we know it has been programmed with the ass) and not very consumer friendly if you are trying something a little bit more extreme then just use the ISP wifi router for your entire home network
for example, they are charging 35€ a month for a static IPv4 instead of a dynamic one and if you want static IPv6, cool for you. (that means, you are screwed)
In fact, I called them today to ask how they were implementing IPv6, their reply was : we don’t do that. (even tough, that ISP was the first one to implement it in Belgium… You know, call centers…)
Anyway, My PPPOE interface is receiving the public /64. I did re build the entire fe80 address plan of my VLAN’s and even played along with a little bit of dhcpv6. I just cannot figure out why it did worked for 5 minutes 3 days ago and how to make that happen again
and for the ping, we are getting 1ms for internal fe80 ping
Here is an interesting fact, I didn’t specify any gateway but every single interface on my router has one specific IPv6 fe80 address that I did not configure and that can only be seen while doing an ip add directly on the router. The web interface won’t see it