I want to use the Linode provided ipv6 pool as public addresses for Docker/Lxc containers. I've tried a few methods now with no success. Networking is not really my skillset I appreciate any help.

After a few attempts I learned that Linode uses custom kernels by default which may or may not respect some ipv6 syctl settings

so I have since changed to a distributed kernel.

Linux li166-218 3.8.0-35-generic #50-Ubuntu SMP Tue Dec 3 01:24:59 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

Docker networking works like this. Docker containers are created and managed by a Docker daemon This daemon upon start will bind itself to a bridged network interface. You can provide a preconfigured interface that Docker will use, otherwise by default it creates its own interface 'docker0' The VMs that docker then creates have their own interfaces which are configured based off the docker bridge.

My current host networking looks like the following. I added 2600:3c01:e000:83::1 from my pool to docker0

docker0   Link encap:Ethernet  HWaddr 56:84:7a:fe:97:99
          inet addr:172.17.42.1  Bcast:0.0.0.0  Mask:255.255.0.0
          inet6 addr: fe80::5484:7aff:fefe:9799/64 Scope:Link
          inet6 addr: 2600:3c01:e000:83::1/64 Scope:Global
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:437 errors:0 dropped:0 overruns:0 frame:0
          TX packets:528 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:427625 (427.6 KB)  TX bytes:50181 (50.1 KB)

eth0      Link encap:Ethernet  HWaddr f2:3c:91:6e:25:63
          inet addr:173.230.156.218  Bcast:173.230.156.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:63829 errors:0 dropped:0 overruns:0 frame:0
          TX packets:40706 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:203827140 (203.8 MB)  TX bytes:5330561 (5.3 MB)
          Interrupt:76

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:53 errors:0 dropped:0 overruns:0 frame:0
          TX packets:53 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:5528 (5.5 KB)  TX bytes:5528 (5.5 KB)

lxcbr0    Link encap:Ethernet  HWaddr 3a:3c:69:7c:80:6c
          inet addr:10.0.3.1  Bcast:10.0.3.255  Mask:255.255.255.0
          inet6 addr: fe80::383c:69ff:fe7c:806c/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:8 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B)  TX bytes:848 (848.0 B)

vethmbiK0d Link encap:Ethernet  HWaddr fe:be:72:ef:1b:61
          inet6 addr: fe80::fcbe:72ff:feef:1b61/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:66 errors:0 dropped:0 overruns:0 frame:0
          TX packets:86 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:9159 (9.1 KB)  TX bytes:9365 (9.3 KB)

I can ping both the ipv6 addresses on eth0 and docker0 from the internet

The network on the VM looks like the following. I added the 2600:3c01:e000:83::2 address to eth0 using lxc config

eth0      Link encap:Ethernet  HWaddr e6:96:c0:8f:d8:32
          inet addr:172.17.0.2  Bcast:0.0.0.0  Mask:255.255.0.0
          inet6 addr: fe80::e496:c0ff:fe8f:d832/64 Scope:Link
          inet6 addr: 2600:3c01:e000:83::2/64 Scope:Global
          UP BROADCAST RUNNING  MTU:1500  Metric:1
          RX packets:51 errors:0 dropped:0 overruns:0 frame:0
          TX packets:40 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:6283 (6.2 KB)  TX bytes:5623 (5.6 KB)

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

I can ping the VM's ipv6 from the host machine. But I can't ping 2600:3c01:e000:83::2 from the internet.

At this point I need help investigating the forwarding to the VM.

I've tried to setup radvd on the bridge interface and enable net.ipv6.conf.all.forwarding. But it seems like anytime I enable net.ipv6.conf.all.forwarding I cant ping my public ipv6 addresses.

i've also tried tweaking

net.ipv6.conf.all.use_tempaddr
net.ipv6.conf.default.use_tempaddr

but if I set them to anything besides 1 I again can't ping my ipv6 address

I'm wondering if i've hit some networking limitation with Linodes Xen setup.

I hand out IPv6 addresses to my Docker containers via radvd just fine. I'm using fd00::/64 so that I can NAT the containers as I do with IPv4, but worked just fine when I was testing it with my public /64. You do need to make sure the following are true:

* You need to have your original IPv6 address up on eth0 (that's the 2600:3c01::dsajdkasdlsajd one, that the /64 is routed to)
* You need to have 2600:3c01:e000:83::1/64 up on the docker0 bridge
* Your radvd.conf should look more or less like this https://github.com/dock0/host_config/bl ... radvd.conf , except with 2600:3c01:e000:83::/64 as the allocation
* You do want net.ipv6.conf.all.forwarding=1 set. You'll need to manually bring up your IPv6 addresses on eth0 and docker0, because enabling this stops them from being autoconfigured

- Les

@akerl Did you have to use a distributed kernel or are you using a Linode provided one?

I use my own kernel because I choose to; Linode's provided kernel works fine with Docker and IPv6 routing.

If you're curious, all my kernel configs live here: https://github.com/akerl/kernels

If you run into kernel-related issues doing what I described, let me know (I'm part of the team that builds the Linode kernels).

- Les

akerl I've tried following your suggestions to no success. I feel i'm missing one detail. It seems as soon as I enable net.ipv6.conf.all.forwarding=1 I can no longer publicly trace-route either of the public ipv6s assigned to docker0 and eth0.

traceroute 2600:3c01::f03c:91ff:fe6e:2563
1   15 ms   1 ms   <1 ms   10ge3-4.core3.fmt2.he.net (2001:470:0:2d::2)
2   7 ms   <1 ms   <1 ms   router3-fmt.linode.com (2001:470:1:3b8::2)

traceroute 2600:3c01:e000:83::1
1   14 ms   <1 ms   <1 ms   10ge3-4.core3.fmt2.he.net (2001:470:0:2d::2)
2   30 ms   4 ms   14 ms   router3-fmt.linode.com (2001:470:1:3b8::2)

Is this an issue? if they are properly routed to the vms should I beable to hit them? This is before I start radvd. When you say:



Unclear what you mean. Should I flush the interfaces and re add the addresses? Is there some autoconfig setting I need to disable?

I also noticed that by default on the machine if I run radvdump there is an existing advertisement for

interface eth0
{
   AdvSendAdvert on;
   # Note: {Min,Max}RtrAdvInterval cannot be obtained with radvdump
   AdvManagedFlag off;
   AdvOtherConfigFlag off;
   AdvReachableTime 0;
   AdvRetransTimer 0;
   AdvCurHopLimit 64;
   AdvDefaultLifetime 90;
   AdvHomeAgentFlag off;
   AdvDefaultPreference medium;
   AdvSourceLLAddress on;
   AdvLinkMTU 1500;

   prefix 2600:3c01::/64
   {
      AdvValidLifetime 2592000;
      AdvPreferredLifetime 604800;
      AdvOnLink on;
      AdvAutonomous on;
      AdvRouterAddr off;
   }; # End of prefix definition

}; # End of interface definition

I had to set accept_ra=2