[rbldnsd] Confused about how to correctly format my zone and startup commands

[Lyle Giese]

Scott Haneda wrote:
> Sorry for the top post, I am on a mobile.
>
> Can I see your zonefile.txt. Obfuscated if needed. I want to use
> rbldnsd as a users database. Sort of a proof of concept to check
> username availability.
>
> I figure this could prove better than hitting a users database and
> dealing with looking up a username in over 20 million records.
>
> Just an idea at this point. Any pointers on managing the zone file are
> appreciated. Currently new additions will be appended with echo 'foo'
> > zonefile.txt.
>
> Removes may use shell commands or a hourly dump of data to replace all.
>
> What is performance on a reload of 20 million records? Does it matter
> to add one records or regen the entire zone? Does order of data in the
> zone play any roll?
>
> Is there any performance hit to pointing named/bind to rbldnsd as the
> rr or should I just talk directly to rbldnsd over a custom port as to
> not interfere with normal port 53 traffic.
>
> I am definstely not clear on the various types and formats of zone
> files I can use. The examples in the docs show use of $ as a replacing
> variable, which does not make sense as to where in a lookup result
> that data is returned. So far I see a 127.0.0.2 result, is that result
> all that can be returned? Can the return ip be changed?
>
> I see a number of DNSBL And WL's returning ranges to give opportunty
> for different subsets of decisions to be made based on the returned ip.
>
> Thanks.
>
> * I also made a macports package portfile for os x to instal rbldnsd
> on any version of os x above 10.3 I believe. Can someone suggest ideal
> binary locations, and what, if any, other files, such as readme's
> should be installed, as well as suggested install locations?
>
> I also want to include a sample zone file and an os x launchd startup
> item, unless this is recommended against.
>
> Who do I contact to make aware of this and get the packages portfile
> linked on the website?

Just for comment here. We have an inhouse RBL plus we rsync a couple of
other outside RBL's for just our internal use. A long time ago, we moved
the 'backend' to a Postgresql database for our inhouse RBL. We manage
all the adds and deletes against the database and once an hour, dump out
a copy of the database and put into a format that rbldnsd likes using
perl. And then we move it into place and rbldnsd reloads it as it
notices the change(hint, don't use cp, only mv).

The text file we feed rbldnsd from our inhouse database is 40MB in size
and we don't notice any issues during a reload. However for redundancy I
do have two servers running rbldnsd and they do not reload at the same
time. I use BIND for our primary DNS and have RR records pointing to our
inhouse RBLDNSD for the lookups.

Yes, you can have something other than 127.0.0.2 as the return value.
Generally mail servers look for not found or an A record. If a mail
server gets an A record, bounce that guy. Things like content filters
and do analyze the return value and make judgements based on the return
values. I only use IP4SET formats.

Lyle