IPv6 testing
-
I just did a gitsync, and got the latest commits - the one your talking about, and rebooted and my HE tunnel came up no problems.
Now it was odd before the reboot after the gitsync I was seeing a crash error on the UI that would not seem to go away. Odd, it had no details in it, close it, send it and it would just pop back up.
But everything seems to be working currently
-
I just did a gitsync, and got the latest commits - the one your talking about, and rebooted and my HE tunnel came up no problems.
Now it was odd before the reboot after the gitsync I was seeing a crash error on the UI that would not seem to go away. Odd, it had no details in it, close it, send it and it would just pop back up.
But everything seems to be working currently
Is your subnet mask 64bits or 128bits for your tunnel? Mines 64-bit
I was able to get it to work after clicking save/apply under the interface configuration page. IPv6 code has had a similar issue a couple of months ago to where we had to manually add a route or click save/apply..
see post http://forum.pfsense.org/index.php/topic,32721.msg179781/topicseen.html#msg179781
-
I'm having to do the same save/apply business. To make matters worse my DSL was in and out all day yesterday and it was leaving the tunnel down every time the DSL came back up. Talk about annoying having to go in and do that every few minutes lol. I finally gave up on it and surfed from bed with my phone lol. Grrrr, seems to be doing it again. In any case my tunnel is a /64. IIRC the direction say to do /128 or /120 but my tunnel wouldn't come up with that config. Since HE's site says /64 for the tunnel I figured it should be /64 on my end as well (same logic as different subnets on a LAN, they won't always talk to each other).
-
Mine is /128 There was some issues with that a while back. You can bring it up /64 and then just change it to /128
-
When I first setup the tunnel we had to use 128 but it has resolve so we can use the 64 subnet that HE assigned to us for the tunnel. After my update yesterday, my tunnel wont come up with a 64 subnet but will come up with 128 subnet… I'm wondering if the code that was removed was a workaround to get the tunnel to come up using 64bit subnet... idk tho...
-
Currently running the 05:55:15 snapshot from 7/15 and whenever my DSL hiccups the v6 WAN still won't come back up by itself, it stays in "gathering data" according to the gateways widget. Obviously I'd love to get the DSL fixed and this probably wouldn't be a problem but alas it's not my house. It affects my grandfather just as much as it does me so you think he would be annoyed by it too. Two days ago when this started he said he would call Bellsouth and see if they can fix it but I have no idea if he did or if anyone ever came out to look at it (normally does this when we get a heavy rain so me things something is attenuating on the line somewhere which granted may not be an easy fix).
Anyway, both my WANv6 and GIF configs are still using a /64. They are the same v6 address space though (7, LAN is 8) so if I change it to a /128 should I change the gif tunnel settings or the WANv6 settings or both? Any chance this will even fix the problem? Note that we do have a static V4 IP from Bellsouth so the IP isn't changing causing this issue.
-
I am using 128 for the gif interface config and 126 on the interfaces page. That comes up reliably after failure for me and reboots.
-
Just changed mine to match, lets see what happens. IIRC after a reboot it was up, it was only when the DSL would hiccup and go offline. DSL/WAN would come back up but it's like it wouldn't reestablish the tunnel. It's gone down several times today but not in the last couple of hours. We'll see.
-
Looks like we're good. DSL just went out, watched the gateways on the dashboard and both came up at the same time just as they are supposed to.
-
Hello everybody,
I just pointed out a little bug and have a feature to request for the DHCPv6 Server.
The bug :
On the DHCPv6 page, when you have a non-common bitmask, showed range is incorrect. My exemple :Address XXXX:XXXX:XXXX:XXXX::1
Subnet mask 65 bits
Available range XXXX:XXXX:XXXX:XXXX:0:0:0:1 - XXXX:XXXX:XXXX:XXXX:ffff:ffff:ffff:ffffThe available range should be :
Available range XXXX:XXXX:XXXX:XXXX:0:0:0:1 - XXXX:XXXX:XXXX:XXXX:7fff:ffff:ffff:ffffand when specifying XXXX:XXXX:XXXX:XXXX:0:0:0:1 - XXXX:XXXX:XXXX:XXXX:ffff:ffff:ffff:ffff to the dhcp server, the server starts but doesn't deliver addresses on the network.
The feature :
You should add a checkbox allowing to disable the "Autonomous address-configuration", setting the rtadvd option "pinfoflags" to 0 (or "") for the interface.
In my case, i need to disable this to keep the router advertisement but not to have autoconfigurated addresses or subnets delivered (because i have several different subnets linked to the same interface).Thanks to everyone for your excellent work on the IPv6 part !
–- EDIT ---
An old bug reappeared :
In line 736, you should change :
$dhcpdv6conf .= " deny unknown clients;\n";by
$dhcpdv6conf .= " deny unknown-clients;\n"; -
I added tickets on the two issues you mentioned.
http://redmine.pfsense.org/issues/1725
http://redmine.pfsense.org/issues/1726The last one, deny unknown-clients is correct in mainline, maybe you're still syncing from Seth's old branch? You don't have the latest code somehow.
-
Thanks to you for adding tickets.
In effect, i hadn't changed the repo URL.
But I changed it now and ran a fresh gitsync from "https://github.com/bsdperimeter/pfsense.git", and the "deny unknow clients" bug is still there.
Am i doing something wrong ? Is this the good URL ?Thanks for your answers !
-
run 'rm -rf /root/pfsense*' and sync again from that URL. That's correct URL, but that's not where you're getting code from.
-
We need to replace the current subnet range calculator for a v6 prefix. Its just outright wrong.
-
I just ran into an interesting possible issue. In osx lion (10.7) all ethernet interfaces receive not one, but multiple v6 addresses. The first address is a stateless autoconfig address which has been the same since the very first pfsense-v6 builds. The second address seems to change every reboot.
I've even had three addresses show in this panel.
I'm not sure if this is a new feature in 10.7, if these addresses were there all along but weren't shown or if there is something up with the dvhpv6 server on my pfsense install. Any thoughts? -
I just ran into an interesting possible issue. In osx lion (10.7) all ethernet interfaces receive not one, but multiple v6 addresses. The first address is a stateless autoconfig address which has been the same since the very first pfsense-v6 builds. The second address seems to change every reboot.
[snip]
I've even had three addresses show in this panel.
I'm not sure if this is a new feature in 10.7, if these addresses were there all along but weren't shown or if there is something up with the dvhpv6 server on my pfsense install. Any thoughts?The same thing happens on Windows, it gets multiple IPs per interface. I haven't looked too deeply, but it would really be nice if it just had the one address, from DHCPv6, and no others.
-
most likely you seeing multiple because of stateless autoconfiguration, not sure how to disable in os x, but on windows you can setup routerdiscovery=disabled with netsh, or you can do with dhcp option as well I do believe.
if you want turn off the auto ipv6 on linux you could do it with say this setting in your /etc/sysctl.conf replacing your interface of course.
net.ipv6.conf.eth0.autoconf=0
-
Is the IPv6 support going to be rolled up into the final release (i.e. SLAAC / DHCPv6)?
EDIT:
OK, so I see that IPv6 is not going to be in 2.0 for release:
http://redmine.pfsense.org/projects/pfsense/roadmap#2.1-IPv6
…so at this point is it better to just sync the git repo to add IPv6 to my current 2.0 snapshot - or start from scratch with the IPv6 ISO's here:
http://files.pfsense.org/jimp/ipv6/
Thanks!
-
Either way is fine. Personally, I just use stock images + gitsync. If you are installing from scratch already, you may as well use an iso from there. Those are getting a bit stale though, I need to make a new set one of these days.
-
@jimp, @johnpoz thanks for the info, I did not know that Windows showed the same behaviour.
I thought it might be an issue, as I had not seen two or three v6 addresses under osx snow leopard (10.6).
Before 10.7, my machines only had the stateless autoconfig addresses.
It seems that 10.7 has silently added a dchpv6 feature, though my dhcpv6 leases page on pfsense doesn't show anything.
This blog post seems to agree with my findings.Also, while searching for a way to disable stateless autoconfig as per @johnpoz's suggestion, I came accross a post on the same site as above explaining how to disable stateless autoconfig on windows xp, vista, 7 and osx 10.7 lion.
