Comment 3 for bug 1737377

Hello Christian,

On 2017-12-11 10:36 AM, ChristianEhrhardt wrote:
> Hi Simon,
> we are currently shuffling around responsibilities for iproute so extra latencies might occur :-/.

I have no urgent need for this. I was simply experimenting with an
IPv6-only lab.

> 2. the Xenial kernel has this type disabled.
> If throwing in an HWE kernel [1] into xenial, does it then work for you?

The required module (ip6_vti) is in 4.4. Loading it and strace'ing the
ip link I got this:

socket(PF_NETLINK, SOCK_RAW|SOCK_CLOEXEC, NETLINK_ROUTE) = 3
setsockopt(3, SOL_SOCKET, SO_SNDBUF, [32768], 4) = 0
setsockopt(3, SOL_SOCKET, SO_RCVBUF, [1048576], 4) = 0
bind(3, {sa_family=AF_NETLINK, pid=0, groups=00000000}, 12) = 0
getsockname(3, {sa_family=AF_NETLINK, pid=1194, groups=00000000}, [12]) = 0
write(2, ""..., 27Unknown tunnel mode "vti6"
) = 27
exit_group(-1) = ?

I get the same on 4.13.0-19-generic from linux-image-virtual-hwe-16.04-edge.

> 1. the iproute package needs fixes that came later than what is in Xenial:
> The package needs to be very close to the kernel version in general.
> Did you have any chance to test the newer 4.9 by forcing it into a Xenial system if that works as well? If you need a ppa for that let me know and I'll create one for you.

As a quick and dirty test, I installed Zesty's debs of iproute2 and
libelf1 on Xenial and that worked with both kernel 4.4 and 4.13.

Thanks,
Simon