[Pdns-users] TinyDNS backend for PowerDNS

[Radek Hladik]

bert hubert napsal(a):
> On Fri, Mar 14, 2008 at 04:56:27PM +0100, Radek Hladik wrote:
>> 	I've noticed the request for TinyDNS backend for PowerDNS in TODO 
>> 	list and I've decided to code something like that. The good news are that 
>> the backend is in beta state and looks working, the bad news are that its 
> 
> Cool Radek! This backend can read tinydns zone files?

Yes, it reads tinydns zone file, parses all that +,=,@,Z,... "tags" into 
respective RRs and servers them via pipe protocol. It can even do AXFR 
(although in not very efficient way)

Only problem is with raw-bytes in some generic records as pipe backend 
get sometimes confused with CR or LF and I did not find a way to escape 
them.

> 
>> 	See http://powerdns.cybersales.cz/ for more details, download, 
>> 	etc... I would like to receive any feedback, ideas, etc... The code still 
>> needs a lot of work and improvements but should be working.
> 
> The main mismatch is indeed that PowerDNS expects backend to not know about
> DNS, not do wildcard processing, not follow CNAME chains etc etc. 

I do not do anything like that. All things I do (locations, timestamps) 
are tinydns specific issues. The only thing I am not sure about are 
wildcards as I do not not know whether PowerDNS will use exactly the 
same wildcard logic as tinydns. In order to mimic tinydns completely I 
could have to implement tinydns wildcard logic in my backend (probably 
as configurable option). See the mentioned page for details.

> 
> Have you measured bulk performance, so how many queries/second you can
> safely do? The dnsreplay program in the powerdns tarball might be of use
> here.

I tried to build this program but it failed with a lot of conflicting 
errors like this:
/usr/include/arpa/nameser.h:282: error: conflicting declaration 
‘ns_t_nxt’ dns.hh:149:
I am using Fedora 7.

I've made simple test with dig and average query time was 2-7ms on my 
old P3 800MHz testing machine. I also measure the PHP lookup itself and 
it is rather quick (something like 0.00023secs - but it is really vague 
number). I've written short performance info on that "project page". The 
main drawback of PHP I see in memory consumption.

> 
> Good luck!
>