DNS error in sporadic situations

[Mathias Körber]

> Hi
>=20
> We use a BIND 8.2.2.P5 running in a Red Hat 6.2 and have a strange=20
> problem. Sporadically, when a user in our site try a URL (any type of=20
> URL, someone highly accessed), the system returns an DNS error for=20
> timeout problems. Well, if the same user try access the same URL=20
> right away after that, the access is normal. The only caracteristic=20
> abnormal is that the problem happens more frequently in a connections=20
> more slow of the site.
> Anyone has any idea about it?

The most likely cause for this is the inability of your caching =
nameserver
to re-fetch the resource record from the domain's authoritative
nameservers, due to network problems, if the old cached entry has
expired.

This may happen
	a) if your own link to the rest of the net is overloaded, or
	   has other reachability problems (eg a route-flap somewhere
	   upstream)

	b) all of the remote site's auth. nameservers have reachability
	   problems. I often see this happening for sites which do not
	   properly locate their authoritative nameservers in diverse portions
	   os the 'net, and such are susceptible to single-points of failure
	   (eg route-flap/congestion near their site etc).

These are the ones I can think of off-hand.

regards

mathias