named coredumping

[Mark Sergeant]

On 21/06/2005, at 1:16 AM, Danny Cooper wrote:

> I am receiving the same problems with FreeBSD 5.4-RELEASE (amd64)  
> bind-9.3.1
>
> DELL PE2850
> 2 x Intel(R) Xeon(TM) CPU 3.20GHz
> 4 GB RAM
>
> named.log.0:19-Jun-2005 11:46:15.945 *** POKED TIMER ***
> named.log.0:19-Jun-2005 18:34:17.314 *** POKED TIMER ***
> named.log.0:20-Jun-2005 02:40:46.193 *** POKED TIMER ***
>
> Depending on the load the of the server it can happen every hour.
>
> In some case's named starts to Max out one CPU and refused to  
> resolve DNS
> until named is killed and restarted.


I've seen the exact same thing on my dual AMD64 caching name server,  
I can only think it is something to do with the Dual CPU's, bind  
9.3.1 and FreeBSD 5.4, even an upgrade to stable as of a week ago  
hasn't fixed it, so for now I have a wrapper script that looks for 99+ 
% cpu usage and then kill -9 and /etc/rc.d/named start . I'd love to  
get to the bottom of it, but know one has  come across this before it  
seems.

Have a look at my posts on heavy named problems in late may.

Cheers,

Mark