[368] in Hesiod
HESIOD type=maildrop class=IN, Sendmail
daemon@ATHENA.MIT.EDU (Russell McOrmond)
Fri Jan 23 00:08:00 1998
From: Russell McOrmond <russell@flora.ottawa.on.ca>
To: hesiod@MIT.EDU, maildomain@flora.org
Date: Fri, 23 Jan 1998 00:06:14 -0500 (EST)
I was bound to find some more data on the question of "Why IN vs HS"
for storing Hesiod information. Here is both a pro, and a stronger CON
for having the information in the 'IN' domain if you will be using the
Sendmail MAP capabilities with libhesiod.
Pro:
I am interested in doing the direct-mapping for as many domains as
possible as I believe that this information can aid in getting email
directly to a user, and allow for more redundancy in Email. In the case
of FLORA, many of the Email addresses are actually 'forwarded' on to
another site. Publishing these forwarded addresses allows 'participating
sites' to send the email direct to the destination.
With the IN class we already have widely distributed NS records, and it
is very easy to know what DNS host to attempt to get Hesiod information
from. If the host didn't have information, it would be a wasted DNS
lookup, but for myself the advantages seemed to overcome that.
Con:
While some have mentioned that the extra bandwidth was a problem, this
was not what convinced me to move my Hesiod information back into the HS
class, and change my config to have classes=HS .
From nslookup:
---cut---
> foo.aracnet.net.
Server: dns1.aracnet.net
Address: 206.222.74.1
*** dns1.aracnet.net can't find foo.aracnet.net.: Server failed
> www.aracnet.net.
Server: dns1.aracnet.net
Address: 206.222.74.1
Non-authoritative answer:
www.aracnet.net canonical name = widow.aracnet.net
Server: dns1.aracnet.net
Address: 206.222.74.1
www.flora.org internet address = 206.248.33.33
flora.org nameserver = nic.flora.ottawa.on.ca
flora.org nameserver = bud.sandelman.ottawa.on.ca
flora.org nameserver = nic2.flora.ottawa.on.ca
nic.flora.ottawa.on.ca internet address = 206.248.33.2
bud.sandelman.ottawa.on.ca internet address = 209.151.24.17
bud.sandelman.ottawa.on.ca internet address = 205.233.54.161
nic2.flora.ottawa.on.ca internet address = 206.248.33.97
> foo.flora.org.
Server: dns1.aracnet.net
Address: 206.222.74.1
*** dns1.aracnet.net can't find foo.flora.org.: Non-existent host/domain
>
---cut---
For some reason, unknown to me at this point, when one looks up an
unknown domain within aracnet.net, a 'server failed' message is
returned. When you ask the same server for an unknown domain within
flora.org, it returns a 'non-existant host/domain' message. There is
something speci al in the way aracnet.net has their DNS configured.
In the case of the Sendmail MAP where Hesiod is looking up in the IN
class, it would fail in it's lookup and queue the message. Eventually
(after 5 days) sendmail would bounce the message back to sender. In a
situation where the IN lookup is not done and instead only an HS class
lookup is done, the message is then delivered to aracnet.net based on the
MX records that it does sucessfully publish in it's DNS.
It seems that the best idea is still to try to build an HS class "root
servers" where those running Hesiod compatible DNS servers and are
publishing information can then tell each other about that. I know I
would be interested to add NS records into my DNS for anyone publishing
'forwarding addresses' using the type 'maildrop'. Alternatively, is
someone else keeping a list available that could be advertised and used?
Note: For those who didn't know, I am using HESIOD with type=maildrop to
do 'email forwarding'. The TXT information is just another Email
address, and it is used directly by a sendmail MAP with some simple
rules. Some initial documentation of what I'm doing is at
http://www.flora.org/flora/server/maildomain.html - I will be updating
this soon to be more complete and up-to-date.
--
Russell McOrmond, Internet Consultant: <http://www.flora.org/russell/work/>
Community Network Comments http://www.flora.org/russell/papers/newyears98.html
Ice storm press & links http://www.flora.org/russell/ice/
First Nations? - Attacked by MAI? http://news.flora.org/flora.perc/340
