Bind crashes my server !

Remi Delon remi at cherrypy.org
Mon Sep 15 11:25:48 UTC 2003


Hi everyone,

I've been having a problem with my slave DNS for several weeks now:

I noticed that sometimes Bind goes wild and start creating more and
more threads/processes until it eventually crashes the server !

After looking in /var/log/messages, I see messages like this:

zone domain.com/IN: refresh: failure trying master 212.34.12.3#53:
timed out

So it looks like for some reason, the slave cannot connect to the
master and maybe it keeps trying and trying, creating more and more
threads/processes.

Most of the time, it *is* able to load the domains from the master, so
this problem only happens once in a while.
On the master DNS, everything always runs fine and I don't see any
error message about Bind in the logs.

On both the slave and master, I'm running bind-9.2.2 that I compiled
from sources myself.
Both servers are running RedHat 7.3.
Oh, and the slave is a dual-processor (I don't see why this would make
a difference though).

So I guess there are two problems here:
1) The fact that the slave sometimes cannot connect to the master to
refresh the domains (this is no big deal)
2) The fact that when that happens, it goes wild and crashes the whole
server. Now, this *is* a big deal !

I just want to get rid of 2)

Does anyone know what might cause Bind to behave like this and how to
prevent it ?

I'm just about ready to write a watchdog just to check that Bind isn't
going wild and to restart it if it is !!

Thanks for your help.

Remi.


More information about the bind-users mailing list