> How do the configured DNS resolvers know to contact the rbl zones?
> Unless it's a stealth slave zone, doesn't seem they would know. I take

They use the information contained in the request from the SMTP proxy, as
well as information gained from authorative servers for the TLDs.

Please do some research on how the DNS system works, or see the URL that
explains this that I posted in a previous email.

> it the current versions no longer have the section under email proxy for
> maps/rbl to specify the server to use? That's what is being implied.

The current versions of the GNATBox software still have the option to
specify what RBLs to use for the SMTP proxy.  The entries are used to
formulate the requests that are sent to the DNS resolver.

> Here you imply that there is still a section under email proxy for
> maps/rbl but this doesn't jive unless you are paying to be a stealth
> slave zone with MAPS or other services that offer this. Since it would
> ask the server(s) configured in that section not the external or
> internal dns. Dnsbl maps type systems and domain name lookups are
> seperate.

There are more options then just running a stealth slave zone.  Some
blacklists do (or did a few years back...I don't have current experience
with those lists.) require that you mirror a copy of their zone and make
all requests of that mirror. (ie: stealth slave zone)

In that case, the recursive server should contain pointers to the
authorative slave server that contains this zone, since you do not want to
use the regular authorative servers.

With dnscache, this would be done by adding a file in
/service/dnscache/root/servers with the name of the zone. The file would
contain the IP address of the server that you wish to contact.  This would
override the normal recursive lookups.)


Most current publicly-accessable RBLs do not require (and in many cases do
not allow) you to transfer a copy of the zone to your authorative servers.
(a "stealth slave zone", if I understand what you mean by that term.)
Instead, they prefer that you do normal recursive lookups using their
servers as authoritative. (that is, the standard DNS lookup routines.)

The SMTP proxy is configured with the subdomain of the RBL.  The proxy
then formulates a request to the RBL based on the source IP address of the
connection.  The request is sent to the resolver, who handles it from that
point.  If the resolver returns any A record, the SMTP proxy will then
request a TXT record.  If the TXT record exists, this will be used in the
error message to the client (and the logs)


> Ahem... If it did look in /etc/resolv.conf then there would be no need
> for this line in sendmail v8.8
> R$-.$-.$-.$- $(host $4.$3.$2.$1.blackholes.mail-abuse.org. $:OK $)
>  or under M4 compatible versions
> FEATURE(dnsbl, `blackholes.mail-abuse.org', `Mail refused')dnl

Yes, there would.  The "host" line is actually a unix command.  This
command is then using the systems resolvers to resolve the request.

ie:

why:/root%# host 2.0.0.127.spews.relays.osirusoft.com
2.0.0.127.spews.relays.osirusoft.com has address 127.0.0.4

If your prefered RBL used a different format, you'd be able to create a
differently formated "host" line to make the request.

The feature you cite only configures the zone that the RBL lookup should
use in the DNS requests, it does not tell sendmail WHO to contact...the
resolver and standard DNS routines take care of that.


> Since we are on the topic of caching/lookups and maps, according to
> mail-abuse.org
>
> "In no case ought you cache the results of a MAPS RBLSM lookup, since a
> blackholed host can right itself and be removed in a matter of seconds."

Probably a good idea.  The authorative server should set a low enough TTL
to prevent caching in most cases.  (some recursive servers are configured
to ignore the TTL and use a standard cache time instead.  AOL does this,
for instance.)

> Also, direct usage via DNS with a mail transfer agent (either sendmail
> or another, which the smtp proxy qualifies as) and subscription via DNS
> are seperate.

Agreed, though the "direct usage" does not mean (necessarily) that the MTA
(sendmail or the GNATBox SMTP Proxy) is making the request directly to the
MAPS servers.  It only means that the MAPS servers are the eventual target
of the query.  The recursive resolvers make the actual direct request to
to the authorative servers for the zone.


> RRset of the zone, so they will never be targets of third party MAPS
> RBLSM (DNS) queries. In order to cause such servers to be queried by
> your mail relays, you must configure the recursive name servers listed
> in your resolv.conf files as zone slaves. (It is normally a bad idea to
> mix authoritative and nonauthoritative data in the same name server, but
> this is a specified exception to that rule.)"

This quote is very BIND-centric, but accurate.   It applies directly to
the GNATBox SMTP proxy, if you choose to be a MAPS secondary server.
You'll need to configure you recursive resolvers to get their data from
the server that contains a copy of the MAPS zone.



---
David Raistrick
        Systems Administrator - Global Technology Associates, Inc
 [EMAIL PROTECTED]

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
To subscribe to the digest version first unsubscribe, then
 e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Archive of the last 1000 messages:
 http://www.mail-archive.com/[email protected]

Reply via email to