[Dnsmasq-discuss] interface-name=<name>, <interface> should also create AAAA record, if available

[Simon Kelley]

On 24/07/13 19:06, Uwe Schindler wrote:
> Hi Simon,
>
> thanks for the information. How does this code sometimes locks up? Is
> it happening when reading the config files or when the network
> adaptor changes its address?

Neither. It's related to handling TCP requests, I think, and results in 
the dnsmasq spinning using all available CPU. I have a configuration 
which can be made to fail (eventually) but I've not been in a position 
to debug it yet.
>
> In any case, I worked around that problem by another "cool trick". As
> told in my earlier mails, providers in Germany change the prefix
> quite often, so the IPv6 address of the router changes quite often.
> This also changes of course the DNS server reported in router
> advertisements and via DHCPv6. The problem I had was that I did not
> want to make the refresh/lease times too short, so I looked for a
> more "static" IPv6 address. The trick I used was:
>
> I assigned an additional "private" and deprecated (preferred lifetime
> = 0) IPv6 address to the network adaptor, that was also not matched
> to the DHCP range template. The Template for constructor:ethX was ::1
> in my case, which did not cover this private address. This private
> IPv6 address was then provided as IPv6 DNS and NTP server in dnsmasq,
> so the clients can rely on a static address (like the private IPv4
> address). I also added a DNS name for this static address in my home
> domain - so interface-name was no longer needed to work with IPv6.
> The "private" address got a preferred lifetime on the network adaptor
> to prevent the kernel to use it as outgoing address. You can still
> listen on it (via ssh/dnsmasq/samba/...) but kernel never uses it to
> start new connections to the outside, so it is really private. The
> clients with public IPv6 addresses can still reach it, as the linux
> box is default gateway and the kernel redirects the packages
> internally to the right interface (which is the same, just different
> IPv6).

That sounds like a sensible solution. IPv6 really is different, and 
there are lots of problems like this which need to be founds and fixed. 
The IETF "homenet" group is working on this sort of thing, and I need to 
get up-to-speed before the meeting in Berlin next week.


Simon.

>
> Uwe
>
> ----- Uwe Schindler H.-H.-Meier-Allee 63, D-28213 Bremen
> http://www.thetaphi.de eMail: uwe at thetaphi.de
>
>
>> -----Original Message----- From:
>> dnsmasq-discuss-bounces at lists.thekelleys.org.uk [mailto:dnsmasq-
>> discuss-bounces at lists.thekelleys.org.uk] On Behalf Of Simon Kelley
>> Sent: Wednesday, July 24, 2013 7:42 PM To:
>> dnsmasq-discuss at lists.thekelleys.org.uk Subject: Re:
>> [Dnsmasq-discuss] interface-name=<name>,<interface>  should also
>> create AAAA record, if available
>>
>> On 16/07/13 11:12, Uwe Schindler wrote:
>>> Hi,
>>>
>>> The config option:
>>>
>>>> interface-name=<name>,<interface>
>>>
>>> should in my opinion also create an AAAA record for the given
>>> interface, if a global ipv6 address is assigned, so the name
>>> returns both A and AAAA records. Would this be hard to do?
>>
>> The code to do it is in the current development branch, but be a
>> bit careful using it: I've seen very rare lock-ups here which I
>> can't yet reproduce to track down the problem.
>>
>>
>> Cheers,
>>
>> Simon.
>>
>> _______________________________________________ Dnsmasq-discuss
>> mailing list Dnsmasq-discuss at lists.thekelleys.org.uk
>> http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
>
>
> _______________________________________________ Dnsmasq-discuss
> mailing list Dnsmasq-discuss at lists.thekelleys.org.uk
> http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
>