"no addrs found for root NS"
Chip Campbell
We have an error message showing up from bind 8.1.1 on our BSD/OS 2.1 and
3.1 systems and from multinet 3.3C on a VMS system (which I think is
equivalent to bind 4.x) relating to our internal false-root setup, which was
recently reworked to move the root zone off to dedicated servers.
On the bind 8.1.1 systems we get this pair of messages approximately hourly.
No functional problems are noted (ie, domain names all seem to resolve ok
for all apps):
Dec 21 14:55:26 our-ns named[11147]: sysquery: no addrs found for root NS
(d.net-root.net)
Dec 21 14:55:27 our-ns named[11147]: sysquery: no addrs found for root NS
(c.net-root.net)
The relevant lines in the conf and cache files are:
zone "." {
type hint;
file "net-root.cache";
};
and
c.net-root.net. 3600000 in a 10.134.16.11
. 3600000 in ns c.net-root.net.
d.net-root.net. 3600000 in a 10.144.73.11
. 3600000 in ns d.net-root.net.
You'll note I swapped the order of the a and ns records compared to the
textbooks in an attempt to fix this. It seems to have helped on some
systems, but on at least three systems the problem persists.
On the VAXes the corresponding error is much more frequent and looks like:
$
%%%%%%%%%%% OPCOM 21-DEC-1998 16:06:05.90 %%%%%%%%%%%
Message from user ADMINS on VAX01
MultiNet Server: ns: ns_req: no address for root server
Worse, we believe it's correlated with a user-level failure to resolve many
hostnames, including the root server names. The config file is
cache . MULTINET:DOMAIN-NAME-SERVICE.CACHE
primary 0.0.127.IN-ADDR.ARPA MULTINET:DOMAIN-NAME-SERVICE.LOCAL
forwarders 10.144.65.11 10.134.1.11 10.130.1.11
and the cache file is
c.net-root.net. 3600000 in a 10.134.16.11
. 3600000 in ns c.net-root.net.
d.net-root.net. 3600000 in a 10.144.75.11
. 3600000 in ns d.net-root.net.
which is identical to the above. All the forwarders named have no problem
resolving anything, though they are producing messages as above (they are
the bsdi systems). But running nslookup pointed at 127.0.0.1 returns "server
failed" for most or all names that ought to resolve fine.
The VAX problems - the message and the resolution failures - subside for an
hour or so immediately after restarting the server, then resume.
Any suggestions or pointers will be appreciated; I really think we've
checked the obvious but if you think otherwise, say so. E-mail or cc'd
postings are welcome.
(The addresses and hostnames in the example have been changed for this
posting to protect the site identity).
Chip
Chip Campbell
Sunnyvale, Calif.
ch...@netcom.com
Barry Margolin
Chip Campbell <ch...@netcom.com> wrote:
>We have an error message showing up from bind 8.1.1 on our BSD/OS 2.1 and
>3.1 systems and from multinet 3.3C on a VMS system (which I think is
>equivalent to bind 4.x) relating to our internal false-root setup, which was
>recently reworked to move the root zone off to dedicated servers.
>
>On the bind 8.1.1 systems we get this pair of messages approximately hourly.
>No functional problems are noted (ie, domain names all seem to resolve ok
>for all apps):
>
>Dec 21 14:55:26 our-ns named[11147]: sysquery: no addrs found for root NS
>(d.net-root.net)
>Dec 21 14:55:27 our-ns named[11147]: sysquery: no addrs found for root NS
>(c.net-root.net)
Remember that the hint file is only used to initialize the root server
information. One of the first things the server does is get the current
root server information from one of the root servers. Once it gets this,
it ignores the information it read from the file. The query it does is
equivalent to "dig . ns @<some root server>". Try this by hand and make
sure it returns all the root servers and includes all their addresses in
the Additional Info section.
--
Barry Margolin, bar...@bbnplanet.com
GTE Internetworking, Powered by BBN, Burlington, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Don't bother cc'ing followups to me.