Greetings list, Bind 8 issue:

Drew J. Weaver drew.weaver at thenap.com
Thu Oct 18 14:03:55 UTC 2001


	Greetings. I have been a lurker on this list for awhile and I've
learned a lot, but everything I've learned has proven useless in solving the
latest problem I've run into. We're a small dial-up ISP, in two of our
cities users cannot send e-mail unless their IP addresses are listed in my
mail servers /etc/hosts file. I've gone the sendmail route with this issue,
and everyone on that side has basically TOLD me it has to be a dns issue but
haven't offered any clues as to what specifically.

Let me give an example

         Say a user dials into our Marietta, OH pop. There are two modem
pools, one of the modem pools has the Class C 206.222.10.0-255 assigned to
it, and a 209.190. IP class assigned to it. If a user is given a 209.190. IP
address everything is absolutely beautiful for them, they have no problem
sending mail, et cetera. If a user gets a 206.222.10.x IP address, they
cannot send mail. What happens is when you make a connection to sendmail
(telnet mail.domain.com 25) it takes almost 3 whole minutes for it to post
the sendmail banner and fire the daemon people have said that this is
because sendmail is doing a lookup on the rDNS for the IP address
206.222.10.x.
         
         206.222.10.x resolves, they all have reverses[I think its worth
noting that they have NEVER had forwards(i.e. a records et cetera) and this
is a recent problem that only started about a week ago] If I put the
206.222.10.x range in /etc/hosts on the mail server it works perfectly, but
obviously that is not a solution it's the sysadmin's duct tape. I've been
mulling over my named.conf and the zones for the set of reverses for the
last few days and I see nothing wrong with any of it. 
         
         Any insight here would be most helpful, thank you for your time and
consideration.
         
         -Drew
         




More information about the bind-users mailing list