ntpd bind() failure: Can't assign requested address

[Damien Fleuriot]

On 12 Mar 2013, at 22:42, "M. Schulte" <m-freebsd at fuglos.org> wrote:

> Hi!
> 
> [First of all, I have posted this question already on the FreeBSD
> forum -- so far without replies -- and now my hope is that the set of
> subscribers here and those of the forum do not completely coincide.]
> 
> I have installed FreeBSD 9.1 on my server (it's a virtualized qemu/kvm
> environment, in case that matters) and during boot the following
> messages appear:
> 
>   ntpd[766]: ntpd 4.2.4p5-a (1)
>   ntpd[767]: bind() fd 23, family AF_INET6, port 123, scope 2, addr fe80::216:36ff:fe74:2076, mcast=0 flags=0x11 fails: Can't assign requested address
>   ntpd[767]: unable to create socket on re0 (3) for fe80::216:36ff:fe74:2076#123
> 
> This happens with a GENERIC kernel. I have not touched any IPv6
> related configuration after install, so everything should be in the
> default state. Neither have I changed the ntp/ntpd configuration.
> 
> Although I could find some threads where people were discussing the
> same problem, none of the mentioned suggestions fixed this for me.
> 
> In particular I tried setting
> 
>   ipv6_activate_all_interfaces="YES"
> 
> in rc.conf -- didn't fix the problem.
> 
> According to ifconfig, my interfaces, re0 in particular, have IPv6
> addresses associated to them. So why would it fail that a process
> tries to bind to it? Note that I'm rather unexperienced wrt IPv6.
> 
> Would be great if somebody could give me a hint into the right
> direction. Anybody else experiencing this? Thank you very much!
> 
> ~ melanie
> 

I'll check tomorrow if we have the same issue at work, although it doesn't look familiar.

We've got a 9.1 with a custom kernel (although nothing related to ipv6 changed) running in KVM.