[Dnsmasq-discuss] dnsmasq does not seem to randomize resolvers

Simon Kelley simon at thekelleys.org.uk
Wed Mar 6 18:02:06 GMT 2013


On 06/03/13 17:32, Sjors Gielen wrote:
> Op 04-03-13 00:29, Ed W schreef:
>> Dnsmasq by default queries all dnsservers simultaneously and locks
>> onto the one which gives the fastest response (rechecking every few
>> queries or every 60 seconds - or some numbers like that)
>>
>> So I guess it's just bad luck that the fastest resolver has a bad
>> record?
>
> Thanks for your response. In my tests the two main resolvers were
> both equally fast (≈ 1 ms as they had the results cached), so even
> though one may be a tiny bit quicker than the other, I doubt it would
> be an explanation for it choosing a specific server 19 out of 20
> times. Though it would be interesting to verify this; is there some
> debug option that shows dnsmasq's train of thought?

The decision is made by which resolver's packet arrives first, so 
microseconds count: I guess the ultimate arbiter is the first common 
switch in the path from the resolvers to the host running dnsmasq.

Of course, one resolver will have an advantage because it gets the query 
sent to it first. The order that the query is sent to the resolvers 
isn't defined, but it's probably a consistent function of their order in 
the configuration file.

>
>> Using strictorder should prove that this is the case?
>
> What do you mean here?

strict-order tell dnsmasq to always send queries to one server at a 
time, in the order they appear in /etc/resolv.conf. Since it defeats the 
ability of dnsmasq to pick the resolvers which are up and fast, it's not 
recommended.


Cheers,

Simon.
>
> Thanks, Sjors
>
>
>
>
> _______________________________________________ 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