How to handle bind 9 as resolver on bad wireless links

[Marc Haber]

On Tue, Aug 21, 2007 at 09:07:04PM +1000, Mark Andrews wrote:
> > On Tue, Aug 21, 2007 at 05:22:44PM +1000, Mark Andrews wrote:
> > > 	Then I doubt that iterative resolution will work either.
> > > 	The only other solution would be to forward to the GPRS's
> > > 	nameservers.
> > 
> > Which would cause the same issue since the GPRS operator's nameservers
> > are not reachable as well during cell change or other connectivity
> > challenges.
> 
> 	But you will have a *local* cache by doing that.  Any queries
> 	for which you already have a answer will get a response.

Yes, that's a clear advantage. I have this with a bind without
forwarders as well.

> > So there is no way to decrease bind's concurrency level and increase
> > the timeouts?
> 
> 	With forwarders (and forward only) you remove all the traffic
> 	except that for the direct queries.  Note the timeouts are
> 	in the stub resolver not named.  Named will try much longer
> 	than the stub resolver.

The current issue is that named is very persistent in retrying and is
monopolizing available bandwidth. I'd like it to be a little more
careful.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 3221 2323190