[Dnsmasq-discuss] Feature request: --server=/*.in-addr.arpa/ overrides --bogus-priv.

[Paul Chambers]

Only tangentially related, but thought it was worth a mention:

http://www.team-cymru.org/Services/Bogons/

They maintain a 'bogon' list of IP ranges that are private (martians) or 
unassigned by the IANA - i.e. IP addresses that are not valid/routable 
on the pubic internet. I currently import it into my netfilter rules 
with a little cron/wget/sed incantation, but it'd be just as easy to 
massage it into something dnsmasq could digest.

-- Paul

Simon Kelley wrote:
> Andrew Rodland wrote:
>   
>> Subject line pretty much says it. I have config lines of the form
>>
>> server=/100.168.192.in-addr.arpa/192.168.100.2
>>
>> (for a VPN "split DNS" configuration), but if the bogus-priv option is enabled
>> these queries are returned NXDOMAIN without forwarding to the upstream server.
>> I've disabled bogus-priv as a workaround, but it's my feeling that an explicit
>> forwarding server should override the general-purpose option. That way I can
>> forward the zones that I know exist somewhere, while still getting the benefit
>> of bogus-priv for completely spurious local IPs.
>>     
>
> This is a fine suggestion, but there are are two reasons not to do it.
>
> 1) It's more difficult to implement than may be apparent, since the 
> bogus-priv logic happens as part of cache processing, and if that yields 
> an address, the forwarding logic which checks the server= lines never 
> gets run.
>
> 2) The current behaviour has been there for a long time, and there's a 
> chance that changing it will adversely affect existing installations.
>
> I'd probably override both these objections and do it anyway, if there 
> wasn't a good workaround, but there is. Do this.
>
> 1) remove bogus-priv
>
> 2) add the lines
>
> local=/.168.192.in-addr.arpa/
> local=/.16.172.in-addr.arpa/
> local=/.10.in-addr.arpa/
> local=/.127.in-addr.arpa/
> local=/.254.169.in-addr.arpa/
>
> instead.
>
> (Your server line will be used in preference to the 168.192 one, since 
> it's more specific.)
>
> HTH
>
> Simon.
>