round-robin

[Barry Margolin]

In article <9seit8$cp5 at pub3.rc.vix.com>,
Joe Kattner  <joe.kattner at adelphia.com> wrote:
>That looks correct. As the DNS & BIND book recommends, shorten the ttl on
>round robin records to something fairly low (they use 60 in the example), so
>that these records don't live very long and get cycled.

That's not necessary.  The caching server caches *all* the records and
round-robins them, so it will still work with a normal TTL.

>The reverse should match the forward, especially in cases where you know
>that a provided service on that host uses forward and reverse name checking
>to make sure they match up as a security feature. You show a 1 PTR to 1 A
>record, in the example which would be correct. 

While the setup he showed will work, I would recommend something like:

$ORIGIN gwu.edu.
smtp  IN A 10.10.10.3
      IN A 10.10.10.4
smtp1 IN A 10.10.10.3
smtp2 IN A 10.10.10.4

$ORIGIN 10.10.10.in-addr.arpa.
3     PTR  smtp1.gwu.edu.
4     PTR  smtp2.gwu.edu.

This way, when you're doing system administration you can go to a specific
server to work on it.  And mailer logs that show where a message came from
will be more specific.

Also, if this hostname is only used in MX records, you don't need the
"smtp" entry.  Just use two equal-preference MX records:

$ORIGIN gwu.edu.
@ IN MX 10 smtp1
  IN MX 10 smtp2 

These MX records will be round-robin'ed, and the sender is also supposed to
try them in a random order.

>-----Original Message-----
>What is the correct way to add pointer records for round-robined records?
>$ORIGIN gwu.edu.
>smtp	IN	A	10.10.10.3
>smtp	IN	A	10.10.10.4
>
>Should I have two pointer records for each IP address pointing to smtp?
>
>$ORIGIN 10.10.10.IN-ADDR.ARPA.
>3	PTR	smtp.gwu.edu.
>4     PTR	smtp.gwu.edu.


-- 
Barry Margolin, barmar at genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.