ipv6 broken: radvd: can't join ipv6-allrouters on <interface>

@havastamas With my build, applying fixes to to get 6RD working, removing spam messaging in RADVD and upgrading to 12.1 code base, my firewall has been up and passing ipv6 tests for over 12 days now.

So I think a solution has been found, its now a matter of priorities. Switching to 12.1 is non-trivial and who knows what other issues that creates. I'm not a networking expert and my application is relatively simple, so all it has done is highlighted that no fatal flaws are out there with starting to move that direction.

w0w

@havastamas
If you are talking about workaround... I see no reason why the workaround would not work, but it’s easier to check yourself whether it works or not.

strangegopher

any updates on this?

w0w

https://redmine.pfsense.org/issues/9577 is updated, but nothing new so far, waiting for the new build based on 12.1

w0w

@jimp when do we expect snapshots based on FreeBSD 12.1? Is it coming early 2020?
Happy holidays!

strangegopher

@w0w said in ipv6 broken: radvd: can't join ipv6-allrouters on <interface>:

@jimp when do we expect snapshots based on FreeBSD 12.1? Is it coming early 2020?
Happy holidays!

im sure upgrading to 12.1 will break other things and require more testing. But I hope this issue will be fixed.

@strangegopher said in ipv6 broken: radvd: can't join ipv6-allrouters on <interface>:

im sure upgrading to 12.1 will break other things and require more testing. But I hope this issue will be fixed.

I’m reasonably current with 12.1 and haven’t noticed any new issues and a few other issues have been fixed. Most of the 2.50 pfsense commits to 12.0 apply to 12.1 without conflicts.

strangegopher

@rschell thats great news! thanks for all the work. Happy holidays!

w0w

@rschell
Ronald, any chances to get already compiled radvd with patch provided on https://redmine.pfsense.org/issues/9577? Just for testing.

@w0w I can give it a try. I’ll have to update my 2.5 version first, it’s been a while.

@w0w said in ipv6 broken: radvd: can't join ipv6-allrouters on <interface>:

@rschell
Ronald, any chances to get already compiled radvd with patch provided on https://redmine.pfsense.org/issues/9577? Just for testing.

I think I got a RADVD version based on 2.5 now. My build machine is based on stable-12 and there was a linker change that required changing one of the linker options (added -z notext).

The test version is based on the 2.5 source changing the behavior of setup_allrouters_membership routine in the RADVD port. From the code listed in 9577, I changed the dimension of the msockets table from 10 to 50 and added a logging message when a socket has been added to the table and also expanded the "can't join" message to include additional information should the IPV6_JOIN_GROUP call fail. Normal operation should log the following messages in the Routing log, one for each interface configured:

Feb 11 08:28:18 	radvd 	78824 	version 2.18 started
Feb 11 08:28:18 	radvd 	78834 	adding ipv6-allrouters on em0, sock: 4, iface->props.if_index:1

I have the txz file ready and have posted it to the redmine issue #9577 for those who wish to test the alternate patch.

w0w

Thanks, testing.

motific

@w0w - any results so far?

kiokoman

i'm start testing right now, we need at least 24 h to tell if it's ok. no prob so far :)

Feb 13 18:58:33 	radvd 	38033 	adding ipv6-allrouters on ix0.100, sock: 4, iface->props.if_index:12
Feb 13 18:58:33 	radvd 	37849 	version 2.18 started

w0w

Yes, I can confirm that IPv6 works good. About 36 hours and no problem.

kiokoman

after all this days it's still working good
great job @rschell

kiokoman

@rschell idk if it's related to this update but lately i had trouble with my ipv6 tunnel, i still need to investigate but the problem is that my tunnel does not work after a reboot
the gif is up but gateway monitor say packet loss 100%
i had to reboot 2 / 3 times in a row to make it work again

@kiokoman Unlikely related to the proposed patch. I'm using an 6RD stf interface and have noticed issues with loss of the WAN connection. Might be a timing issue with the gif interface, not certain.

kiokoman

some kind of race condition at boot maybe, also if i wait some hours the tunnel start to work alone. but it's faster if i reboot until it start working. idk the problem started last week just after i moved pfsense from bare metal to esxi, probably something to do with that

@kiokoman I've noticed that bare metal vs Hyper V behave differently in subtle ways.