listen-on failing

[Barry Margolin]

In article <sch1qc2ibi730 at news.supernews.com>,
 <phil-news-nospam at ipal.net> wrote:
>Being down to one server for a short period, I had to move a bunch of
>IP addresses all onto this one server.  That seems to be messing up
>bind as it now takes from 2 to 10 minutes just to get up to answering
>queries.  I noticed a number of error messages about it thinking that
>another name server was listing to some of the IP addresses (this was
>not true) and deleting those interfaces.  So in the hopes that fewer
>IP addresses for it to listen to will speed things up, and perhaps
>eliminate the race condition that may be the cause of the problems,
>I tried to configure 5 IPs (the ones actually used for DNS) in the
>listen-on option.  However, it seemed to reject 4 of them, so I am
>back to it taking 2 to 10 minutes (random) to start up.  Here's the
>syslog messages I got:
>
>Mar  9 22:43:00 mira named[31372]: /etc/named.conf:30: cannot redefine
>listen-on for port 53
>Mar  9 22:43:00 mira named[31372]: /etc/named.conf:32: cannot redefine
>listen-on for port 53
>Mar  9 22:43:00 mira named[31372]: /etc/named.conf:34: cannot redefine
>listen-on for port 53
>Mar  9 22:43:00 mira named[31372]: /etc/named.conf:36: cannot redefine
>listen-on for port 53
>
>Can someone tell me what the exact meaning of this is?

I think you have multiple listen-on statements, which is not right.  You
should have a single listen-on statement that lists all the IP addresses:

	listen-on port 53 { 1.1.1.1; 2.2.2.2; 3.3.3.3; ... };

>Also, I get these messages:
>
>Mar  9 22:01:29 mira named[23378]: There may be a name server already
>running on [127.0.0.1].53
>Mar  9 22:01:29 mira named[23378]: deleting interface [127.0.0.1].53
>Mar  9 22:03:29 mira named[23378]: There may be a name server already
>running on [206.97.151.201].53
>Mar  9 22:03:29 mira named[23378]: deleting interface [206.97.151.201].53
>
>There is no other name server running.  The previous name server process
>was shutdown and then killall was used to be sure everything was cleared
>out.  What can I do to make bind NOT delete them?

This could be a consequence of the above problem.

>Until I get the other server back online, I cannot be shutting down bind
>and restarting it all the time to get interfaces back that were deleted.
>I need to have bind absolutely and positively listening on 127.0.0.1 and
>the 5 other IPs I tried to configure.
>
>Is there an ndc command to tell bind to restart a listen-on interface
>that it had deleted, without shutting down and restarting everthing
>(in which case it will probably delete some other interface after
>waiting 2 to 10 minutes to get going).

"ndc reconfig" should make it reread named.conf and update the interfaces
it's listening on accordingly.

-- 
Barry Margolin, barmar at bbnplanet.com
GTE Internetworking, Powered by BBN, Burlington, 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.