UDP packet loss

Thomas Schulz schulz at adi.com
Fri Jun 9 17:54:56 UTC 2006


In article <e6bglg$ric$1 at sf1.isc.org>,
bert hubert  <bert.hubert at netherlabs.nl> wrote:
>On Thu, Jun 08, 2006 at 04:01:18PM +0200, Pavel Urban wrote:
>> I've tried to find out the cause of some strange resolving problem we 
>> are experiencing. From time to time, our server seems to ignore some 
>> requests, while others are answered promptly. It seems to me that the 
>> problem is UDP packet loss.
>
>The symptom is UDP packet loss, the problem is that the nameserver is not
>processing UDP packets quickly enough. You can raise buffers but this merely
>hides the underlying problem, and also causes new ones as the nameserver
>will be answering old & stale questions.
>
>I suggest you look at your system load to see if your nameserver is keeping
>up.
>
>A good way to see this is by running vmstat 1 and checking if there is any
>'id' capacity left.

Something that I have not seen mentioned yet.  Bind tries to cache a lot
of data in memory.  If there is not enough memory, it may start to
page/swap.  If it does that, the performance will be terrible.  That
could cause bind to fall behind and the queries to pile up.  Check on
how much paging/swapping your system is doing.

>
>-- 
>http://www.PowerDNS.com      Open source, database driven DNS Software 
>http://netherlabs.nl              Open and Closed source services
>
-- 
Tom Schulz
schulz at adi.com



More information about the bind-users mailing list