[Dnsmasq-discuss] Cache Time Threshold

Simon Kelley simon at thekelleys.org.uk
Mon Apr 14 13:37:49 BST 2008


Robert Diamond wrote:
> On Wed, 09 Apr 2008 19:36:20 +0100
> Simon Kelley <simon at thekelleys.org.uk> wrote:
> 
>> My take on this is that the DNS admin set the Time-to-live, and it 
>> shouldn't be over-ridden. If the DNS admin wanted a longer time, they 
>> would have set it. Even if you get away with extending the TTL for
>> most domains, eventually you'll find one which breaks.
> 
> dnsmasq could have a "keep-alive" cache, which updates cache entries
> as they expire for a maximum time. Obviously set to something low,
> 1-10 keep-alive cache entries should be sufficient for most uses,
> without ever using an expired ttl. ie: "--keep-alive-cache-count=3" and
> "--keep-alive-cache-time=3600" would keep the last 3 cached dns
> requests up to date for up to 1 hour past the last request for that
> entry.
> 

That's an intriguing idea: the downside is the extra DNS load. Unless
there's some clever manual or automatic way to distiguish between
domains that you're interested in and "other" then you example of 3
domains won't get very far - OTOH keeping everything alive will possibly
use more cache, and generate more DNS traffic.

Maybe there's a place for a daemon which is configured with a list of
domains, and does DNS queries on each just in time to keep the cache
valid? It would need a DNS cache to work, but would otherwise be
completely independent of it.

Cheers,

Simon.




More information about the Dnsmasq-discuss mailing list