Can't restart BIND.

Now BIND server shutting down...

Trying to stop BIND server... done
Trying to start BIND server...
ERROR while trying to start BIND server
Check the error reason, fix and try again

Every time I try to restart BIND, I get this error in the log:

Oct 25 17:28:59 localhost named[2994]: app.c:561: unexpected error:
Oct 25 17:28:59 localhost named[2994]: isc_app_shutdown() pthread_kill: No such process

Anyone know how to fix this?

I'm having the exact same problem on redhat, I have spent a good 45 minutes now searching google and every place I could find. The most I can come up with is this uncomprehensible translation from a russian newsgroup:

URL

From what I could understand out of this translation I assume redhat's init script for named is not working properly, or, for some reason the PID being recorded is not correct. In either case I don't know how to solve this. I personally use Debian and ran into this while trying to fix an unrelated problem on a friend's redhat server.

So far I have not found a solution. Please let me know if you do manage to figure this out or have any further clues.

Thanks.

What version of BIND are you running, and tell me more information on the circumstances where it happens. Like, what does ALL the named messages say in the syslog file. I need to know this cause I'm looking at the BIND source, and I found where that error msg is posted to the log, but I need to know from where the function is called.

I am running BIND 9.2.1, on redhat 9.

This occurs when I try to use 'rndc stop'. It fails to stop named, and the process still lives afterward. I can use rndc status, reload, and other such commands.

BIND is otherwise functional. The only way for me to stop the process is to kill it.

While searching google I came across a few other people experiencing the same problem, oddly enough, all seemed to be using redhat 9, and all the information I found was fairly recent, within the past few months. I suspect this is a redhat specific problem, my Debian setup is quite similar and works fine.

When I start named (using /etc/init.d/named start) I see the following in my syslog:

Dec 4 11:34:24 ns named[25200]: starting BIND 9.2.1 -u named
Dec 4 11:34:24 ns named[25200]: using 1 CPU
Dec 4 11:34:24 ns named[25200]: loading configuration from '/etc/named.conf'
Dec 4 11:34:24 ns named[25200]: no IPv6 interfaces found
Dec 4 11:34:24 ns named[25200]: listening on IPv4 interface lo, 127.0.0.1#53
Dec 4 11:34:24 ns named[25200]: listening on IPv4 interface eth0, xxx.yyy.zzz.195#53
Dec 4 11:34:24 ns named: named startup succeeded
Dec 4 11:34:24 ns named[25200]: listening on IPv4 interface eth0:1, xxx.yyy.zzz.245#53
Dec 4 11:34:24 ns named[25200]: listening on IPv4 interface eth0:2, xxx.yyy.zzz.246#53
Dec 4 11:34:24 ns named[25200]: listening on IPv4 interface eth0:3, xxx.yyy.zzz.247#53
Dec 4 11:34:24 ns named[25200]: command channel listening on 127.0.0.1#953
Dec 4 11:34:24 ns named[25200]: zone 0.0.127.in-addr.arpa/IN: loaded serial 1997022700
Dec 4 11:34:24 ns named[25200]: zone foo1.tld/IN: loaded serial 2003110717
Dec 4 11:34:24 ns named[25200]: zone foo2.tld/IN: loaded serial 2003110500
Dec 4 11:34:24 ns named[25200]: zone foo3.tld/IN: loaded serial 2003111500
Dec 4 11:34:24 ns named[25200]: zone localhost/IN: loaded serial 42
Dec 4 11:34:24 ns named[25200]: zone foo.tld/IN: loaded serial 2003110702
Dec 4 11:34:24 ns named[25200]: zone ns.foo.tld/IN: loaded serial 1068052662
Dec 4 11:34:24 ns named[25200]: zone ns2.foo.tld/IN: loaded serial 1068188761
Dec 4 11:34:24 ns named[25200]: running
Dec 4 11:34:24 ns named[25200]: zone foo1.tld/IN: sending notifies (serial 2003110717)
Dec 4 11:34:24 ns named[25200]: zone foo2.tld/IN: sending notifies (serial 2003110500)
Dec 4 11:34:24 ns named[25200]: zone foo3.tld/IN: sending notifies (serial 2003111500)
Dec 4 11:34:24 ns named[25200]: zone foo.tld/IN: sending notifies (serial 2003110702)
Dec 4 11:34:24 ns named[25200]: zone ns2.foo.tld/IN: sending notifies (serial 1068188761)

(note I replaced some hostnames and IPs for security reasons, but the general structure and format are unchanged)

With named running and operating normally, I tried executing 'rndc stop', the named process still lived, so I checked my syslog :

Dec 4 11:32:15 ns named[22718]: app.c:561: unexpected error:
Dec 4 11:32:15 ns named[22718]: isc_app_shutdown() pthread_kill: No such process

Is there some way I can get more verbose output of the error?

I don't know of any way to have more verbose output.

But I've done some snooping around. To spare you a bunch of programmer talk, just know that when BIND is compiled, it checks on the availability of LINUXTHREADS and performs accordingly. This is said about Redhat 9 however:



For more details consult page 7 on the following document:

http://www.redhat.com/docs/manuals/...es-x86-en-9.pdf

I don't know if BIND deviates from the POSIX standard, since I'm not familiar with the POSIX standard and don't care to be. But at least I gave you an educated guess of a reason for the error.

I recommend that you upgrade BIND. If the problem persists contact ISC about the bug. Do me a favor and let me know how things go.

Yeah, that sounds like it might be the problem. I did a little digging and found this:

URL

notice the bug report is for the same problem, and one of the comments contains an interesting reference to linuxthreads and NPTL.

Thanks for your help!

yeah buddy, they went way farther than I did in determining the problem and solution. But they didn't say if BIND 9.2.3 still has the problem. If you upgrade, let me know if it does.

Just did a redhat kernel upgrade, from 2.4.20-8 to 2.4.20-24.9, which fixed the problem. I can now stop named using rndc without any problems.

So anyone experiencing this problem should make sure they have an up to date kernel.

Thanks again for your help.

I happened to stumble on this playing with DNS UPDATE

You can raise the debug level using rndc.

Increment debugging level by one.
rndc trace

Change the debugging level.
rndc trace level

Set debugging level to 0.
rndc notrace