So I am running multi-WAN, and I do have NPt set up to translate my /48 with an HE.net /48 on my backup connection, and I see the same behavior you posted a screen shot of. I tried removing the NPt rule and still observed the same behavior on my primary WAN (IPv6 address being reported as the router WAN IP, not my desktop IPv6 address)

Any suggestions on things I could check or this just a side effect of using multi-WAN and gateways w/fail-over?

@ksdehoff Odd that yours didn't work for the static mapping, maybe because I enumerated the entire interface ID (::7d86:e96:bb0c:fe85 for example). So I don't have to mess with changing anything in the static mapping. I had another issue unrelated to it (caused by Snort of all things) and I had as part of troubleshooting, unchecked the 'do not allow release...' setting and rebooted, and yep the prefix changed and the servers got new IPs with the same interface ID and the new prefix. So I am happy with that small victory.

@nineeyes

maybe you have a wrong address somewhere
or maybe the address changed

I would check the $FRIEDL_FARM interface settings to see if there are wrong settings there

Hmm... look... another ISP (in Germany this time) with the same issue. I guess Comcast isn't the only one broken. Can this be looked into now to see where the problem lies as far as pfSense's handling of prefix size received being different from prefix size requested?

https://forum.netgate.com/topic/159463/ipv6-not-working-wan-and-lan-interface-getting-an-ip-adress-not-any-client

@gamienator-0

At the moment, you're using an even number prefix. What happens if you pick an odd one? That /63 moves the boundary between the prefix and suffix over by one to the left. Will a prefix ID 1 now be the same as 0?

@jknott Yes, the "routing" is done by link local address.

@johnpoz

Perhaps a touch. However, I have noticed a lot of misunderstanding about IPv6, because people are so used to IPv4. While many things work the same way, some others are quite different. When I had that IPv6 problem, a couple of years ago, I found I had to educate the 2nd level tech support (I wouldn't waste my time with 1st) and senior tech at my ISP on the finer details of how some things worked with IPv6.

As for the WAN address, a public address is entirely optional with IPv6, relying on the link local address for routing. That seems to be quite a leap for many to understand.

@johnpoz said in ipv4 only no more:

How many vlans do you have?

I have 8 subnets/vlans.

LAN (default vlan): Switchs, APs and controller
Infrastructure: Pi-Hole
Home: iPhones, iPads, Macbooks
Media: LG TV, Roku TV, Apple TVs, Sonos Speakers
Server: Synology and QNAP NAS
Printer: HP printers
IoT: Kindles and Bike Computers
Guest:

I have 3 SSIDS

Freeside: Enterprise Radius assigned VLAN
Chiba: PSK Radius assigned VLAN by MAC address
Sprawl: Guest

I put everything I could on Freeside, including one of my printers that supports WPA2 Enterprise EAP-TLS. Lots of fun with Apple Configurator for the others.

Chiba gets the kindles, bike computers and Roku TV. Before anyone has a fit, no you can't get on this network by MAC address only. They are only used to do VLAN assignments. You still have to know the pre shared key. Unifi is kinda misleading with this, they call it 'RADUIS MAC AUTHENTICATION". I tested this and found that you have to have a user in Radius that matches the MAC address and the PSK. Radius shows it as a successful logon if you have no password or the wrong password but the AP doesn't connect you in that case. Maybe you could do this on an open network or do something in Radius to make it a MAC bypass. That is a terrible idea.

Sprawl is the guest network.

Everything that is stationary is on a wired connection with the exception of the Roku TV and one Apple TV.

One printer (an all-in-one) is on a cart and connects to Freeside (didn't know it supported Enterprise EAP-TLS until recently, never bothered to look when I bought it) :)

I violate the F out of the L2 segregation using avahi (mDNS/Bonjour) and udpbroadcastrelay (SSDP, for the Sonos). I'd post up all my firewall rules but that would just serve to make me look dumber than I already do. They get the job done but are not nearly as locked down as they could be.

There is a lot that could be improved. We're probably going to move late spring/early summer and that will be the time to get some gear that is quieter and more energy efficient. A Netgate appliance and new switch(s). Get rid of my unifi stuff and replace them with Ruckus APs if I can find some for a decent price used. Put bigger drives in my Synology and retire the QNAP. There's always something...

@jan_berg This approach seemed to be working for me: https://wiki.cable-wiki.xyz/OPNsense

Caveats:

Can't be done through UI, needs to be executed in a shell. The tunnel will not be visible in the UI. Doesn't persist. Would need to re-execute every time the WAN comes up and has a global IPv6 assigned. Need to extract the AFTR name and its IPv6 address. In my case, the name comes through via DHCPv6 from the ISP as option 64. Could extract it via tcpdump. Then resolved it to an IP address and used that when setting up the tunnel. Breaks again if AFTR name/IP changes.

So, no real DS-Lite support in pfSense currently, but possible to set up manually.

@jknott Honestly, I don't think I ever intentionally set anything up for that (nor knew it was an option to disable it either). It's just something that's always been there on the dashboard. I assumed it was pfsense pinging the gateway address and getting the answer (since the gateway is usually given by dhcp on the WAN).

I just found the checkboxes to disable it - all good :-)

8be775e9-6ceb-4d0d-90d3-7915e64cb8fa-image.png

@pfguy2018

I recently changed the rules for my guest WiFi VLAN to IPv6. in some cases it was only necessary to change from IPv4 to IPv4 & IPv6. I have only one rule that is IPv6 specific and none that are IPv4 specific. That IPv6 one is to block anything within my prefix.

Here are my rules:

Screenshot_20201212_161304.png

@jknott OK - so have sussed it -am on a pure IPv6 connected PC now! So Static IPv6 address on link, DHCPv6 disabled, but RA set as assisted with a DNS server with the link ipv6 address set on the RA tab.

So I think this is SLAAC + RDNSS working properly?

Even managed to use a literal IPv6 address for the pfsense box - https://[ipv6 address] needed in edge - square brackets eh?

Irony of testing though one of the Test -Ipv6 sites I was using didn't resolve an IPv6 address (test-ipv6.com) where as ipv-test.com was happy!

@bob-dig

All the addresses appear automagically. One of each type is consistent, based on the MAC address. The privacy addresses are based on random numbers. The only thing I configure is the DNS entries, which I point to the consistent addresses. I do not ever use a privacy address for DNS, as it would only last for a week. It is also possible to have consistent addresses based on a random number, for those who are worried about someone tracking their MAC address.

@gertjan

While I haven't seen 2 link local addresses in a device with only 1 interface, multiple routeable addresses are common. For example, this computer, once it's been up for a week, will have a total of 16 routeable addresses, 8 global and 8 unique local. Of those, one of each is consistent and MAC based and the others are privacy addresses, of which I get new ones every day, with them expiring after 7 days.

I see. Will tracerouting the ipv4 addresses shown in the registration process be sufficient to tell if a specific tunnel endpoint is a good choice or will it require registration and bringing up the tunnel itself to be sure?

True, line 9 :
@kaj said in ICMPv6 Trouble?:

prefix ::101:101/128 {

is not ok at all.

@wlp94611 see https://redmine.pfsense.org/issues/10984

You've probably either figured this out already or just decided to ignore it but I have found that those errors are typically caused by an IPv6 client that doesn't support DHCPv6 and your IPv6 Router Advertisements are configured not to support SLAAC (set to either "Managed" or "Disabled" on that interface).

Under "Services/DHCPv6 Server & RA", change your RA mode to "Assisted" or "Stateless" on the interface those clients are connected to and this error should disappear.

@AmC_OldSarge said in Android Clients unable to reach Internal Exchange Server:

From what I am reading, by default, android uses IPv6 for DNS.

Most things prefer IPv6 by default, but if it's not available they'll go immediately for IPv4.

Yep the Livebox is in it's box since the day I received it, never opened the parcel 🤣

For now, from pfSense WEB UI, I have to put each LAN in "Static IPv6" and give it a 2a01:aaaa:bbbb:<prefix ID>::/64 subnet by hand...

So you are right, tomorrow if the prefix changes... everything will break !

To solve that I was thinking the "Track Interface" feature will do the job but it seems to be broken when using DHCP6C in "Advanced Configuration" on the WAN side...

@cnrd said in pfSense does not reply to NS sent by ISP router, ISP does not respond to DHCPv6 request as a result:

As stated here: https://www.freebsd.org/security/advisories/FreeBSD-SA-08:10.nd6.asc

The solution described below causes IPv6 Neighbor Discovery Neighbor Solicitation messages from non-neighbors to be ignored.
This can be re-enabled if required by setting the newly added net.inet6.icmp6.nd6_onlink_ns_rfc4861 sysctl to a non-zero value.

I think a package coming from a global address to a link local would be considered a non-neighbor.

Here is what I read on Redit:

"II. Problem Description

IPv6 routers may allow "on-link" IPv6 nodes to create and update the
router's neighbor cache and forwarding information. A malicious IPv6 node sharing a common router but on a different physical segment from another node may be able to spoof Neighbor Discovery messages, allowing it to update router information for the victim node."

Now, take a look at the packet containing the neighbour solicitation or advertisement and check the hop limit. It is 255. This is protection against that threat as a router would have to decrement it from 0, but a 0 hop limit would cause the packet to be dropped. This guarantees the packet originated on the local network. If it's any other number, the packet originated elsewhere and with a hop limit other than 255 or 0.

@signalz

If one device gets an IPv6 address but another doesn't, you have a local problem. Likely you have pfsense misconfigured.

As for AAAA records, there is no way for any DNS server to know what name you assign to an address, unless you configure it. I have both the pfsense DNS server and a public server available. I put the host names I choose on those servers. Also, bear in mind, there are consistent addresses and privacy addresses, which change every day. Point the DNS to the consistent addresses. Consistent addresses are often based on the MAC address, but may be based on a random number. Either way, you point the DNS to the consistent address. Also, you will probably have a WAN IPv6 address, which is likely not used for routing.

So, when you determine what your consistent addresses are, enter names for them in the pfsense DNS server.

@evberd thanks for posting. I just lost connectivity again today, triggering the multiple-reboot cycle of pFsense until it would finally pull an IPv6 for my WAN and LAN. While I can grit my teeth and suffer through this until I get it going before the next disruption, it is truly vexing behavior. I woke up to no connectivity a few days ago due to Comcast equipment maintenance, and had to spend half an hour and was late to work because I needed to get the internet going for distance learning for my kids.

That said, I've read what you've written a couple times, and it hasn't completely sunk in. I have tried using VLAN before and failed miserably. I am not an expert.

I do use dynamic DNS as Comcast has changed my IP address several times over the past 6 months, I think due to equipment upgrades which is not a bad thing. The performance is impressive. Dynamic DNS does actually work, and does not seem to be a terrible solution for residential service.

DHCPv4 can use a "client identifier" such as a DUID, just like DHCPv6. (rfc4361)

I suspect this is what you are seeing.

@qwerty123
I've got Spectrum and it's the same thing- I have some homelab servers that I don't want dynamic but I have implemented ipv6 internally (dual stack). What I did, right or wrong, was set the WAN interface to DHCP6, and the LAN and 2 VLANs to track, specifying prefix IDs for the VLANs (/56 hint worked). I also set the WAN to NOT send a release to the ISP under DHCP6 Client Config. Instead of using the DHCP6 from my servers, I set that up in PFSense and setup static IPV6 mappings for the servers. But instead of specifying the prefix which we know may change, I set the addresses as ::{interface identifier} hoping that if I don't specify a prefix, and it changes, at least the prefix ID and the interface ID remain as I specified and my servers still get a valid routable IPV6 ip. When it changes I will have to update a couple of ALIASES that list those networks, and some settings under Services/DHCPv6 Server & RA. Hopefully I won't have to mess with the static mappings though.

Still some things to examine there but yes the ndp output is what I was looking for.

@Bob-Dig said in Network Prefix Translation (NPt) Failing:

First, JKnott is always right.

That's not what my ex says! 😉

@skilledinept

Forget about the management interface and radius for a moment. The purpose of an AP is to provide a layer 2 connection between devices. That's it. So, it should pass IPv4 & IPv6 equally well. I get the impression your radius server is IPv4 only. Is that a problem? Can the clients only connect via IPv6? These days, dual stack is quite common and normally IPv6 is preferred, with fallback to IPv4.

Really?

After another call to my ISP the problem finally solved!!!
There was nothing from my side!

Thank you all for your support.

@Jim-Coogan what ended up being a show stopper for me was my ISP only allocating a /64 range to my modem. For DHCPv6 relay to work with pfSense acting as a router you need to be able to use DHCPv6 with Prefix Delegation. To do that you need a bigger allocation than /64, e.g. /56 or /60 etc.

The bridged interfaces are ipv4 only. the issue is that the automatically generated rules are floating rules so they apply globally

Yeah as Jknott has stated its normal.. It is also possible that the wan never even gets global address, and just use link-local.. Im not a fan of that, I like to see a global address on my wan ;)

There is plenty of IPs to go around ;) ISP can afford to assign a global to the transit network ;)

edit: to expand on the sheer number of IPs.. A min assignment from RIR for an ISP is a /32 - I just got one for a ipv6 project we are doing from Arin.. That is 65K /48s ;) or 4 billion /64s..

Comcast got a /9 - which is 36 quadrillion /64s - you would think they can afford a few /64 for transit networks ;)

And its not like they can't get more... It took a couple of weeks to get the /32 - all you have to do is show basic need.. And a basic plan on how your going to use them..

These ISPs telling users they can only get 1 /64 is just nuts... You can head over to HE and they will give you a /48 you can tunnel for free. I have had mine for over 10 years..

ISPs should have no issues giving users either a 48 or at min a /56 and using a global for their transit network.

What if you use a remote receiver?

From experience I've learned that DDNS in pfSense, or any other appliance only works when you're "gentle" to it, meaning one hostname. As soon as you add additional hostnames, domains it'll fail to update them so I got a VPS and installed pfSense on it for the static IP they give you, starting at USD5/mo, sometimes less, it's the cheapest static address you can rent.

I actually use it for the GIF to HurricaneElectric and tunnel both IPv4 and IPv6 to the local pfSense, I have about the same latency in local IPv4 as in tunneled IPv4 and (double-) tunneled IPv6.

But where I'm going with this is:

I also use my remote instance's public address as the monitor IP for the local WAN gateway. And since I can contact the remote instance locally through the tunnel, I get statistics on it with any tool, like from which IP a tunnel is has been brought up--which I know would only be mine. "Loopback" Stats. This data can be sent to a syslog server or queried through SNMP. You can query all sorts of data, I check consumption because the VPS has a data cap, I'm used to not having it because of my ISP so this is really handy, you can set it to notify you through a bot on Telegram, Matrix, classic email or a buttload of other integrations it has:

Screen_Shot_2020-10-09_at_14_37_41.png
Screen_Shot_2020-10-09_at_14_51_00.png

The first one is LibreNMS, completely free, does SNMP and syslog, you don't need scripts or databases because it's meant to keep historic data, it's all there as long as you feed it. The second one is VMware's vRealize Log Insight, also free (*with purchase) it only does syslog but it's very comprehensive, king of syslogs, it can proxy the syslog to yet more servers and has this thing called agents, custom-made-on-site apps preconfigured to send data to it and reconfigurable remotely. It's very cool.

LibreNMS is like a 2core/2G/20G VM if I'm not mistaken, Log Insight is much hungrier but you can tweak it before first starting it, I discovered. Both need fast disks.

@virgiliomi said in IPv6 ping/traceroute from Windows 10 PCs:

Before I try to offer a reason, can I make a guess that you have Verizon FiOS? 🙂

Edit: Never mind... I see you're not from the US. We have an ISP in part of the US that has an ICMP Traceroute issue (affects only Windows, not Linux/Mac since they use UDP instead of ICMP by default). I thought that might have been what you were experiencing.