Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

disappearing MX records?

34 views
Skip to first unread message

John Kerbawy

unread,
Dec 28, 1999, 3:00:00 AM12/28/99
to
Hello. Over the past few days I've been having some mail problems because
my MX records keep disappearing. Once, I went away for the weekend, and
when I came back, I didn't have mail I was supposed to have, and there
wasn't a MX record propogated out on the 'net. I had to do something like
restart BIND before it worked again. This has happened more than once.

I run BIND8 on a OpenBSD box with a DSL connection. I don't understand
only 1 of the 2 domains breaks. The box that does the DNS is the primary
NS for the domain that doesn't break (belluccis.com), and is the secondary
domain for the domain that does break (maKintosh.com). It just so happens
that the box doing the master NS for maKintosh.com has been down for
around a week now. Is this the cause of my problems and if so, how do I
fix it? Below is what I think is relevant from my named.conf

options {
directory "/";
named-xfer "/named-xfer";
};

logging {
channel logfile {
file "log/named.log" size 20m;
severity info;
print-category yes;
print-severity yes;
};

controls {
unix "/var/run/ndc" perm 0600 owner 0 group 0;
};

<snip>

// Master zones

zone "belluccis.com" IN {
type master;
file "master/belluccis.com.zone";
};

// Slave zones

zone "maKintosh.com" IN {
type slave;
file "slave/maKintosh.com.zone";
masters { 206.160.152.252; };
};

Thank you.

__
.j0hn [jo...@maKintosh.com]

Jim Reid

unread,
Dec 28, 1999, 3:00:00 AM12/28/99
to
>>>>> "John" == John Kerbawy <jo...@NOC.maKintosh.com> writes:

John> I run BIND8 on a OpenBSD box with a DSL connection. I don't
John> understand only 1 of the 2 domains breaks. The box that does
John> the DNS is the primary NS for the domain that doesn't break
John> (belluccis.com), and is the secondary domain for the domain
John> that does break (maKintosh.com). It just so happens that the
John> box doing the master NS for maKintosh.com has been down for
John> around a week now. Is this the cause of my problems and if
John> so, how do I fix it? Below is what I think is relevant from
John> my named.conf

John> options {
John> directory "/";
John> named-xfer "/named-xfer";
John> };

Here's one problem. You've told the name server to run an executable
called /named-xfer when it needs to retrieve a zone with a zone
transfer. [ie when it fetches the data for any slave (secondary)
zones.] Did you really install named-xfer in the root directory?
Since there's probably nothing exectuable called named-xfer in /, the
zone transfer attempts will fail. These failures should be reported in
the name server logs.

There should also be messages in the logs about the non-availability
of the master server for makintosh.com. If that box is down, there's
nothing you can do about it apart from complain to the administrator
of that system. You can of course fix the strange choice of location
for named-xfer in the options{} statement above while you're waiting
for the master name server for makintosh.com to come back.

It's also a bit strange to make the name server use "/" as its working
directory. You might not want it to dump core in the root filesystem.
It's usually a good idea to store all the name server zone files
in one place and making this the name server's working directory. This
should be simpler to manage than sprinkling the files across various
directories in the root filesystem. If you're happy with the existing
set-up, then carry on. You're the one that manages it after all.


Barry Margolin

unread,
Dec 28, 1999, 3:00:00 AM12/28/99
to
In article <Pine.BSO.4.10.991228...@NOC.maKintosh.com>,

John Kerbawy <jo...@NOC.maKintosh.com> wrote:
>Hello. Over the past few days I've been having some mail problems because
>my MX records keep disappearing. Once, I went away for the weekend, and
>when I came back, I didn't have mail I was supposed to have, and there
>wasn't a MX record propogated out on the 'net. I had to do something like
>restart BIND before it worked again. This has happened more than once.
>
>I run BIND8 on a OpenBSD box with a DSL connection. I don't understand
>only 1 of the 2 domains breaks. The box that does the DNS is the primary
>NS for the domain that doesn't break (belluccis.com), and is the secondary
>domain for the domain that does break (maKintosh.com). It just so happens
>that the box doing the master NS for maKintosh.com has been down for
>around a week now. Is this the cause of my problems and if so, how do I
>fix it? Below is what I think is relevant from my named.conf

I'm not currently getting any response from I.maKintosh.com -- it doesn't
respond to DNS queries or pings.

The fact that your master was down for a week is quite significant. You
have the Expire time of the domain set to 1 week:

makintosh.com. 86400 SOA NOC.makintosh.com. john.NOC.makintosh.com. (
1999122800 ; serial
28800 ; refresh (8 hours)
7200 ; retry (2 hours)
604800 ; expire (7 days)
86400 ) ; minimum (1 day)

So after a week of failing refreshes, the slave server will expire the zone.

--
Barry Margolin, bar...@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.


0 new messages