[Dnsmasq-discuss] verification of how RA works.
Simon Kelley
simon at thekelleys.org.uk
Mon Nov 26 19:36:26 GMT 2012
On 26/11/12 18:00, Gene Czarcinski wrote:
> On 11/26/2012 11:49 AM, Simon Kelley wrote:
>> On 26/11/12 16:08, Gene Czarcinski wrote:
>>> I believe that RA will work in the manner described below but have not
>>> verified/tested this yet.
>>>
>>> If there are 4 IPv6 (gateway) addresses specified on an interface and
>>> one of those addresses has a dhcp-range specified, then having a
>>> dhcp-range=<ipv6-addr, ra-only specified for the other three address
>>> will result in:
>>>
>>> 1. state-full RA for the IPv6 subnet with the dhcp-range specified.
>>>
>>> 2. state-less RA for the other three subnetworks.
>>>
>>> Yes, it is not clear to me why one would define multiple subnets on the
>>> same "fabric" but that is a capability in libvirt for virtual network
>>> interfaces.
>>>
>> Please test it. I'm feeling cautious today, so I'll go as far as to say
>> that what you describe is how it should work, it's how I'd expect it to
>> work, and if it doesn't work like that it's a bug. It's quite likely
>> that no-one has ever done the test before, so please try it.
>>
> In order to test this I am going to need to do a little coding but I was
> planning to do this anyway.
>
> Interesting. As far as I can tell, if dnsmasq works as described above,
> it does something that radvd cannot do (at least as a single instance
> with a single configuration file). In radvd, the Managed Flag is
> specified for the interface, not the subnetwork.
OK, you prompted me to look at the code, which makes radvd's behaviour
more understandable. The Managed flag is in the header of the
route-advertisement packet so it has, logically, to apply at all the
prefixes contained therein. The dnsmasq implementation sets the managed
flag if any of the prefixes has DHCPv6 available, but clients will take
is applying to them all.
If that causes clients to solict DHCPv6 for all the prefixes, then only
the one with a suitable dhcp range will get an answer.
Cheers,
Simon.
>
> Gene
>
> _______________________________________________
> Dnsmasq-discuss mailing list
> Dnsmasq-discuss at lists.thekelleys.org.uk
> http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
>
More information about the Dnsmasq-discuss
mailing list