Hello,
Brian told me that I should use the email robot, but
we would like to use
the
hamnetdb.net portal to manage our sites, hosts and links, so maybe it
would be convenient to export DNS data from there automatically and
periodically.
hamnetdb.net has a DNS data exporter tool, so we can easily
load the exported data to our DNS server instead of writing a script which
periodically does the data conversion and mailing to the AMPR robot.
That’s partly true.
Yes, the changes in the DNS set are quite fast visible.
We have two hosts which are good connected to the internet and are part of the internet.
Thus we started the delegation concept as a test, and it works well.
That delegation works with forward-resolving (i.e.
db0xxx.subregion.de.ampr.org).
For backward-resolving, the PTR requests should have also to be delegated. We don’t do
that (see below, complete state at
ucsd.edu).
Since I’ve seen so many projects in the years that startet with good ideas and sooner or
later, they lingered around and the person who was responsible was not reachable anymore.
Ok, we in germany have built up a team and do our job for long years now.
Nevertheless, imagine if you have one DNS server, the server goes down, nobody has a copy
- then your complete pr/hamnet resolving is offline until you get fixed it.
Here comes our (heavy!) scripting into play.
In germany we had, since the old packet-radio time, a concept, where our regions have the
absolute control over their subnet (
subregion.de.ampr.org). We have a few masters (called
Hub West, South, East, ..) which get notifications (-> zone transfer) from the regions
nearby; they notified each other, and on one of these systems our script checks for the
changes for
ucsd.edu. [
xxx.]<callsign>.<region>.de.ampr.org is mapped to
[
xxx.]callsign.ampr.org.
This concept was also transferred for our
asXXX.de.ampr.org notation for the HAMNET.
The script does many consistency checks. DNS entries have to contain callsigns.
We disallow things like
pmr.ampr.org,
pn.ampr.org,
yogi-gate.ampr.org, etc..).
If a forward-resolving for a [
xxx.]call.ampr.org entry exists in the
ucsd.edu database AND
[
xxx.]call.<region>.de.ampr.org has disappeared, and delete it.
If a new entry appears, then add it (but only if it is consistent, that means, [xxx.]call
does not exist (that means: does not point to another address).
=> Every call points to _one_ IN A address. And every PTR backpoints to exactly one
forward-address (=> no flapping names on traceroute). The PTR points back to
[
xxx.]callsign.ampr.org.
Looking at the hamnetdb project: it made things easier for the SysOPs. They plan and
document their network, and the result is a complete dns-set for the region. It’s exported
to one of our three DNS hubs (or a regional DNS server) and is almost suddenly visible
(=> resolving of [
xxx.]<callsign>.<region>.de.ampr.org - from both, hamnet
AND the internet). [One limitation: if a region has decided to have a NS delegation to
ns.db0xxx.subregion.de.ampr.org, and this host is HAMNET-only, then from the internet that
addresses could not be resolved].
On another DNS hub, where our script runs, we regulary (on a daily basis) look for changes
and do the mapping to [
xxx.]<callsign>.ampr.orgpr.org.
Imagine if our team would break (we sit in the same train on accident, earthquake, etc).
Then
ucsd.edu still knows everything about our network in germany, and changes could still
be made manually to the amprdns-bot.
About CNAME and MX records: we submit manually to
ucsd.edu after request.
vy 73,
- Thomas dl9sau
> Maybe
> the use of
hamnetdb.net is the reason why Germany and others use own DNS
> servers.
>
> --
> Varga Norbert
>
http://www.nonoo.hu/
>