named[353]: sysquery: no addrs found for root NS ..........

[Ion-Mihai Tetcu]

On Wed, 8 Sep 2004 07:28:11 +0900
horio shoichi <bugsgrief at bugsgrief.net> wrote:

> On Tue, 7 Sep 2004 14:24:37 +0300
> "Toomas Aas" <toomas.aas at raad.tartu.ee> wrote:
> > Hi!
> > 
> > > >From time to time I get this:
> > > 
> > > Sep  7 12:57:44 it named[353]: sysquery: no addrs found for root NS (d.root-servers.net)
> > > Sep  7 12:57:44 it named[353]: sysquery: no addrs found for root NS (a.root-servers.net)
> > > Sep  7 12:57:44 it named[353]: sysquery: no addrs found for root NS (c.root-servers.net)
> > > Sep  7 12:57:45 it named[353]: sysquery: no addrs found for root NS (h.root-servers.net)
> > > Sep  7 12:57:45 it named[353]: sysquery: no addrs found for root NS (f.root-servers.net)
> > > Sep  7 12:57:45 it named[353]: sysquery: no addrs found for root NS (b.root-servers.net)
> > > 
> > 
> > This problem plagued me for a long time on several FreeBSD 4 servers 
> > running BIND 8 from the base system. Google finds numerous discussions 
> > on this problem in various lists/newsgroups but a solution is rarely 
> > offered.
> > 
> > Finally, I found someone's theory in a NetBSD (or was it OpenBSD) 
> > forum. I can't tell whether it is true or not, but it makes sense 
> > to me.
> > 
> > If your BIND is configured to use a forwarder and this forwarder is 
> > really good then BIND (almost) never needs to contact the root servers. 
> > The root zone times out in memory and it is not reloaded from disk. It 
> > is only loaded when BIND is started. Thus, if your BIND finally needs 
> > to contact a root name server after a long time of getting all 
> > responses from forwarder, it turns out that the data for root zone is 
> > not available...
> > 
> > Now, as I said, I cannot tell whether this theory is true or not. What 
> > I can say is that on all 4 machines where I run BIND I configured 
> > one of two workarounds:
> > - use "forward only" so you *never* need to check the root zone
> > - do not use forwarders at all so you check the root zone fairly 
> >   frequently.
> > 
> > I did this almost a year ago, and after that I never have had this 
> > problem again. HTH.
> 
> Hmm, then the easiest cure against OP's would be
> periodically (say, per week) requesting purposely
> wrong request (e.g., nslookup example.heh) ? 

Actually the uptime was 3 days (I've upgraded to 5.3B3) and I'm _sure_
to have entered some non-existent addresses in my web browser.
 


-- 
IOnut
Unregistered ;) FreeBSD "user"
5.3-BETA3 - try `sysctl debug.witness_watch=0`
and prepare to fly :-)