CRM-SPAM: Re: 1 second delay - more detail and specifics

Tim Peiffer peiffer at umn.edu
Mon May 30 23:09:50 UTC 2005


Wayne,

Find out if the server is even running by checking the process tables.  
Pay particular attention to the process size vs what is resident in 
core.  Historically, I have found that the only time I had issues was 
during a reload/restart period on a machine with insufficient memory to 
keep bind resident, alternatively when I was dumping or calculating 
statistics.  You can signal the running named process with -USR1 to turn 
on debug, or stop the nameserver and restart with the debug level you 
want on the command line ( -d5 ).

Try
kill -USR1 `cat /var/run/named.pid`
a few times to bump up the debug level and then reload..  See what the 
server says.
kill -USR2 `cat /var/run/named.pid`
to turn it off.

For dig, debug using something that is authoritative on your server.

  dig -v @127.0.0.1 localhost. +norecurse

Regards,
Tim Peiffer
University of Minnesota.

Wayne Walker wrote:

>On Mon, May 30, 2005 at 12:09:26PM -0400, Barry Margolin wrote:
>  
>
>>In article <d7fb5m$1fhv$1 at sf1.isc.org>,
>> Wayne Walker <wwalker at bybent.com> wrote:
>>
>>Try using the -v option to host to see what it's doing.
>>    
>>
>Very useful!  It's getting the answer from the _second_ DNS server in
>resolv.conf.  Since there was no error, I assumed the answer was
>compinmg from the first server.  For some reason, the bing running
>locally isn't handling the request and the resolver is rolling over.
>
>No errors in messages... :(
>
>dig reports that the server is not responding, but it is running and not
>generating errors.
>
>  
>



More information about the bind-users mailing list