IPv6 beta for Bigpipe

Forums › Spark New Zealand (including Skinny and BigPipe) › IPv6 beta for Bigpipe

1 | ... | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | ... | 29

/dev/ttys0
11024 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#1759983 10-Apr-2017 06:58

Brumfondl:

michaelmurfy:

carlwithaK Yes your configuration is looking right. One thing I would change is the MTU as if you're using UFB you can have a MTU of 1508 on Ethernet interfaces and 1500 on the PPPoE interface.

Out of curiosity, why those values for the MTU? I thought BigPipe used 1492.

Doing so takes care of the PPPoE overhead and because BigPipe support it on UFB :)

Michael Murphy | https://murfy.nz | https://keybase.io/michaelmurfy - Referral Links: Sharesies | Electric Kiwi
Are you happy with what you get from Geekzone? Please consider supporting us by making a donation.

NordVPN

Affiliate link

Affiliate link: NordVPN allows you to securely access the Internet, encrypt your connection and keep your browsing history private.

IPv6pipe

98 posts

Master Geek

Trusted

BigPipe

#1760011 10-Apr-2017 09:24

Can confirm 1508 byte packets are supported on Bigpipe UFB. :)

carlwithaK

14 posts

Geek

#1760421 10-Apr-2017 20:26

It's taken a while to find a window to repeatedly drop the internet connection, but I've since rebooted the ER-X at least a half dozen times with no effect. I assume the MTU thing is performance only and not related (specifically) to IPv6?

As there is nothing obvious wrong in the setup I might take it over to the Ubiquiti forums to see if there is anything ER-X specific that I have missed. I'll update the result either way.

carl

carlwithaK

14 posts

Geek

#1760479 10-Apr-2017 22:15

I've just noticed that I do not have the /etc/radvd.conf file mentioned above, though at some point previously I was definitely able to run "release dhcpv6-pd interface pppoe0" without the error relating to this file. When I get a change I will try to start again and do a bunch of reboots without trying too many configurations.

michaelmurfy

/dev/ttys0
11024 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#1760528 11-Apr-2017 02:22

@carlwithaK flick me a DM. It'll be something simple but I'm willing to take a look via SSH if you like?

Taubin

516 posts

Ultimate Geek

ID Verified

Subscriber

#1762268 12-Apr-2017 15:12

I'll post this here as well, just in case it is related to BigPipe and not something on my end.

I have two windows 10 computers, a desktop hard wired, and a surface pro 4. The desktop is running Win10 enterprise, and the laptop Win10 Pro. About every 36-48 hours, they both stop routing over ipv6 until I disable and re-enable the adapter or reboot the computer completely. Both will show as having 2 ipv6 addresses, a temporary and a full address. Attempting to browse to any site that is strictly ipv6 (ipv6.google.com) fails entirely.

There is a mac laptop as well as two android devices (phone and tablet) that do not have this issue at all. They keep routing through ipv6 with no problem.

My setup is a drytek 310 in bridged mode to an edgerouter lite, set up using michaelmurfy's tutorial for both the erl setup and the ipv6 setup. The desktop is wired, and the laptop is wireless through a tp-link in ap only mode.

I noticed today, that my desktop dropped ipv6 when it was given a second temp ipv6 address. This happens on both machines. I noticed a few hours prior to this happening each time, sites start preferring ipv4 over 6 on these machines (for example right now my surface is routing geekzone through ipv4 instead of 6). I suspect it will drop entirely within a few hours on this machine. I posted on the ipv6 subreddit and it was suggested that the blocks given by BigPipe are being reassigned every 36-48 hours and not being picked up by my windows machines. I'm not sure if that's the case or if it's something I've stuffed up, but I'm open to any and all suggestions.

ZL2TOY/ZL1DMP

MrTomato

139 posts

Master Geek

Lifetime subscriber

#1762271 12-Apr-2017 15:27

Taubin:

I have two windows 10 computers, a desktop hard wired, and a surface pro 4. The desktop is running Win10 enterprise, and the laptop Win10 Pro. About every 36-48 hours, they both stop routing over ipv6 until I disable and re-enable the adapter or reboot the computer completely. Both will show as having 2 ipv6 addresses, a temporary and a full address. Attempting to browse to any site that is strictly ipv6 (ipv6.google.com) fails entirely.

It sounds like you might be experiencing a separate issue, but it's worth pointing out that Windows 10 1607 (Anniversary Update) is known to have problems with DHCPv6. Despite numerous reports (including from me), incredulous that they'd let something like this slip through the net and then just ignore it, Microsoft finally started paying attention when it stymied their own efforts!

I believe the problem's fixed in 1703 (the new Creators Update), if you're feeling adventurous / masochistic.

Taubin

516 posts

Ultimate Geek

ID Verified

Subscriber

#1762273 12-Apr-2017 15:31

MrTomato:

I believe the problem's fixed in 1703 (the new Creators Update), if you're feeling adventurous / masochistic.

Thanks for the link, sad they ignored it. I'll toss the creators update on my surface as there's nothing important on there and see if it corrects the issue. Cheers!

ZL2TOY/ZL1DMP

SirHumphreyAppleby

1988 posts

Uber Geek

#1762278 12-Apr-2017 15:37

Taubin:

I have two windows 10 computers, a desktop hard wired, and a surface pro 4. The desktop is running Win10 enterprise, and the laptop Win10 Pro. About every 36-48 hours, they both stop routing over ipv6 until I disable and re-enable the adapter or reboot the computer completely. Both will show as having 2 ipv6 addresses, a temporary and a full address. Attempting to browse to any site that is strictly ipv6 (ipv6.google.com) fails entirely.

...

I posted on the ipv6 subreddit and it was suggested that the blocks given by BigPipe are being reassigned every 36-48 hours and not being picked up by my windows machines. I'm not sure if that's the case or if it's something I've stuffed up, but I'm open to any and all suggestions.

You should be able to confirm that the permanent and temporary addresses are within the same /64 (the first 64 bits of the address should not change). Temporary addresses are normal when using SLAAC. You could disable this feature (using the netsh command), or set static addresses, and see if connectivity fails without the address changes.

I've been using IPv6 and SLAAC for about six years without any issues.

Taubin

516 posts

Ultimate Geek

ID Verified

Subscriber

#1762285 12-Apr-2017 15:49

I take that back, my surface is already on 1703 and has been for a good while now. I had enrolled it in the insider program right when I got it. Unfortunately it seems they haven't fixed it in the update. I'll wait for it to be resolved by them. Thanks for the info, I had tried searching but didn't see it in my efforts. Cheers.

ZL2TOY/ZL1DMP

SirHumphreyAppleby

1988 posts

Uber Geek

#1762290 12-Apr-2017 15:57

Taubin:

I take that back, my surface is already on 1703 and has been for a good while now. I had enrolled it in the insider program right when I got it. Unfortunately it seems they haven't fixed it in the update. I'll wait for it to be resolved by them. Thanks for the info, I had tried searching but didn't see it in my efforts. Cheers.

I suggest having a look at the output of ipconfig /all before and after the issue occurs to see which DNS servers are assigned. The issue MrTomato has linked to, while appearing perfectly valid, shouldn't stop IPv6 connectivity from working.

Taubin

516 posts

Ultimate Geek

ID Verified

Subscriber

#1762292 12-Apr-2017 16:05

The dns servers haven't changed when it happened. That was the first thing I checked (it's always dns!). First server is my pihole and the second is google. Both for ipv4 and ipv6.

ZL2TOY/ZL1DMP

SirHumphreyAppleby

1988 posts

Uber Geek

#1762313 12-Apr-2017 16:47

Taubin:

The dns servers haven't changed when it happened. That was the first thing I checked (it's always dns!). First server is my pihole and the second is google. Both for ipv4 and ipv6.

Recently, my Let's Encrypt certificates failed to update on my Web server... some idiot (me), forgot to set the DNS server. Oops.

Were able to resolve IPv6 hostnames when the problem occurred?

I would use ping -6 initially. Internally, ping would use getaddrinfo() or the Windows API equivalent. This will return the addresses Windows resolves the hostname to. After that, I would move on to nslookup, and sequentially query each server to see if IPv6 addresses can be resolved. This will return the results of queries made directly to the servers.

Obviously, if IPv6 connectivity is somehow broken, you won't be able to query the public IPv6 DNS servers, but if it's a DNS issue, which may be related to the DHCPv6 DNS bug MrTomato mentioned, you could work around the problem by not providing the IPv6 DNS server addresses to hosts on your network.

Taubin

516 posts

Ultimate Geek

ID Verified

Subscriber

#1762318 12-Apr-2017 16:54

I'm not able to ping them at all using ping -6 website.tld when it goes down on the machines. I have the surface set to the dns servers directly, on the desktop they are set to the router. nslookup only gives the ipv4 address when ipv6 goes down on the machines.

ZL2TOY/ZL1DMP

carlwithaK

14 posts

Geek

#1763543 13-Apr-2017 07:47

@michaelmurfy sorry for the slow reply and thank you for the offer. I'll be on touch to see if we can figure this out.

It was suggested that I should be setting the prefix-only setting, as other similar ISPs use link local addressing on the WAN. Is this the case for BigPipe?

1 | ... | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | ... | 29