"/etc/init.d/named reload" not doing much [SOLVED]

[Nahor]

Hi,

I have bind (9.2.2-r3, chrooted) running fine. But when I update my zone files, "named reload" doesn't do much.
In the logs, I see it reloading:

Code: Select all

Aug  5 17:05:13 server named[5804]: loading configuration from '/etc/bind/named.conf'

So rndc communicate correctly with named.
But when I query:

Code: Select all

# dig foobar.domain.com @localhost

; <<>> DiG 9.2.3 <<>> foobar.domain.com @localhost
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 62121
;; flags: qr aa rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0

;; QUESTION SECTION:
;foobar.domain.com.             IN      A

;; AUTHORITY SECTION:
domain.com.             86400   IN      SOA     server.domain.com. hostmaster.domain.com. 2004072603 28800 7200 2419200 86400

;; Query time: 0 msec
;; SERVER: 127.0.0.1#53(localhost)
;; WHEN: Thu Aug  5 17:06:07 2004
;; MSG SIZE  rcvd: 96

Which shows the old serial number. My zone file contains:

Code: Select all

domain.com              IN SOA  server.domain.com. hostmaster.domain.com. (
                                2004080500 ; serial
                                28800      ; refresh (8 hours)
                                7200       ; retry (2 hours)
                                2419200    ; expire (4 weeks)
                                86400      ; minimum (1 day)
                                )

If I restart bind instead of reloading it, it is then correctly updated.

Any idea?

[adaptr]

I don't understand what you think each component does.
named is the actual nameserver daemon - it doesn't accept arguments.
rndc, on the other hand, is the remote nameserver daemon controller - that is what you use to send commands to the running nameserver.
So the correct command would be:

Code: Select all

rndc reload

and watch the magic happen...

[Nahor]

Well, "named" is also the script in /etc/init.d that managed the named daemon. This script has also a "reload" function that actually calls "rndc reload".
And I did run "rndc reload" manually too but it behaves exactly the same.

I also tried:

Code: Select all

kill -SIGHUP `pidof named`

but same thing, I get the syslog message saying that named is loading the configuration file but the serial number returned by dig is still the old one.

[kashani]

I did a few tests on my side and it works perfectly for me. When doing the reload you should see the following in the logs.

Code: Select all

Aug 06 08:54:19.063 loading configuration from '/etc/bind/named.conf'
Aug 06 08:54:19.070 zone domain.com/IN: loaded serial 2004080503

The fact that you don't see the domain show up in the logs is puzzling... I'd suspect it's caused by your chroot enviroment, but couldn't say for sure or how to fix it.

kashani

[adaptr]

Well, "named" is also the script in /etc/init.d that managed the named daemon. This script has also a "reload" function that actually calls "rndc reload".
And I did run "rndc reload" manually too but it behaves exactly the same.

Oops
Color me noobie, then - I thought you meant the actual executable.

[Nahor]

Juste a quick update:

I got a new server so I was able to experiment a bit without breaking my corworkers DNS. And I found the culprit. My named.conf file looked like this:

Code: Select all

[...]

controls {
        inet 127.0.0.1 allow { localhost; } keys { "server-key"; };
};

key "server-key" {
        algorithm hmac-md5;
        secret "...";
};

[...]

zone "localhost" IN {
        type master;
        file "pri/localhost.zone";
        allow-update { key "server-key"; };
        notify no;
};

[...]

Removing the "allow-update" line fixed the problem.