query throtlling in bind9

[Ladislav Vobr]

reposted again, anybody can clarify this?
..................


Can somebody shed some light on the way bind9 is retrying the
nameservers, in case of all the nameservers are unreachable for
particular domain.

Does it send for every single recursive request to this domain, several
retries to each of servers all the time? (this I believe was bind8
behavior, it sent for every recursive requests at least 3 retry packets
to each of the servers again and again)

or

Does it send only single request to each of the nameservers, and after
it times out it sends another one and again and again? (this would mean
that the will be just one packet in 90 seconds to each of unreachable
servers, regardless the traffic coming from the recursive clients be it
hundreds such a requests per second for example)

or

does it slow down with the time? (understanding that there is no reason
to retry with the same frequency, if the domain is down for let's say
one year:-)...

thanks for any response

Ladislav