[Pdns-users] Re: Some PowerDNS Recursor oddities

[bert hubert]

On Tue, May 16, 2006 at 03:20:53PM +0200, Stephane Bortzmeyer wrote:
> If you receive a A record with a TTL of zero, do you increase it
> because you believe you are "more right" than the zone owner (I often
> thinks so but I thought I was the only one to be bold enough).

704	 	            //      if(rr.ttl < 5) 
705	 	            //  rr.ttl=60; 

These lines were in PowerDNS, commented out. I'm unsure if I should offer
this feature. It does help perceived and real performance.

> > Also, what does it *mean* to have a lower TTL than specified in the
> > parent zone? It means that either:
> 
> > 1) the parent zone is making false statements about the expected
> > validity of its answers
> 
> > 2) You are making changes faster than the registry can follow in any
> > case
> 
> No, it means that I know no DNS registry where the user can specify a
> TTL for the NS delegation. All database schemas I've seen for DNS
> registries make the TTL a global variable. Protocols like EPP or RRP
> do not allow to transmit a TTL.

The configuration where the TTLs differ stil does not "mean" anything
useful, regardless of whether registries offer you to configure it so.

There are lots of scenarios where the TTL as specified in the parent zone
will end up in caches, you can't rely on having a lower TTL in the auth zone
"meaning" anything.

But still, I've updated PowerDNS now, but this has not helped performance
one bit, nor has it fixed any problems.

DNS as a protocol really sucks.

Kind regards,

bert hubert

-- 
http://www.PowerDNS.com      Open source, database driven DNS Software 
http://netherlabs.nl              Open and Closed source services