8.2.1 ndc restart trouble

jlewis at lewis.org jlewis at lewis.org
Mon Oct 11 15:26:49 UTC 1999


I recently upgraded a system that's a master DNS server for a few hundred
zones to Red Hat 6.1 which comes with Bind 8.2.1.  Since then, we've been
having some trouble with named when our DNS admin does an ndc restart.

Oct 11 10:55:13 gsvlfl-ns-1 named[3235]: named shutting down
Oct 11 10:55:13 gsvlfl-ns-1 named[3235]: USAGE [clipped]
Oct 11 10:55:13 gsvlfl-ns-1 named[3235]: NSTATS [clipped]
Oct 11 10:55:13 gsvlfl-ns-1 named[3235]: XSTATS [clipped]
Oct 11 10:55:14 gsvlfl-ns-1 named[11292]: starting.  named 8.2.1 Fri Sep
24 14:52:24 EDT 1999 
^Iroot at porky.devel.redhat.com:/usr/src/bs/BUILD/bind-8.2.1/src/bin
/named
Oct 11 10:55:14 gsvlfl-ns-1 named[11292]: hint zone "" (IN) loaded (serial 0)
Oct 11 10:55:14 gsvlfl-ns-1 named[11292]: Zone "atlantic.net" (file
master/atlantic.net): No default TTL set using SOA minimum instead
[it goes on to load many more zones here]
Oct 11 10:55:46 gsvlfl-ns-1 named[11292]: master zone "rigelcorp.com" (IN)
loaded (serial 1999100400)
Oct 11 10:57:46 gsvlfl-ns-1 named[11292]: There may be a name server
already running on [127.0.0.1].53
Oct 11 10:57:46 gsvlfl-ns-1 named[11292]: deleting interface
[127.0.0.1].53
Oct 11 10:58:46 gsvlfl-ns-1 named[11292]: listening on [209.208.0.2].53
(eth0)
Oct 11 10:58:46 gsvlfl-ns-1 named[11292]: Forwarding source address is
[0.0.0.0].1042
Oct 11 10:58:46 gsvlfl-ns-1 named[11335]: Ready to answer queries.
Oct 11 10:58:46 gsvlfl-ns-1 named[11335]: ns_req: 
sendto([209.208.0.33].1475): Connection refused
Oct 11 10:58:46 gsvlfl-ns-1 named[11335]: ns_req:
sendto([209.208.0.27].1037): Connection refused
Oct 11 10:58:46 gsvlfl-ns-1 named[11335]: ns_req:
sendto([209.208.42.132].1390): Connection refused
Oct 11 10:58:47 gsvlfl-ns-1 named[11335]: ns_req:
sendto([204.144.0.252].1048): Connection refused


At this point, named is running, but it's not answering requests.  Is it
possible that doing an 'ndc restart', and then doing another while named
is still sending/receiving NOTIFY messages could cause the new named to be
run before the old one is fully dead, resulting in the new one not being
able to listen on the same interfaces/ports?...or is this likely a bind or
perhaps linux kernel problem?  The kernel in use is the 2.2.12 that comes
with RH 6.1.  For now, the DNS admin has been told to use ndc reload
instead, which he should have been doing before anyway.

Also, is there any way to get bind 8.2.1 to not complain about "No default
TTL" for every zone?  Since 8.2.1 is new to me, and causing problems, I'm
hesitant to add a TTL statement to every zone since I'll likely downgrade
to 8.1.2 if 8.2.1 keeps causing trouble.
 
----------------------------------------------------------------------
 Jon Lewis *jlewis at lewis.org*|  Spammers will be winnuked or 
 System Administrator        |  nestea'd...whatever it takes
 Atlantic Net                |  to get the job done.
_________http://www.lewis.org/~jlewis/pgp for PGP public key__________



More information about the bind-workers mailing list