named stops responding on zone transfers

Jim Reid jim at rfc1035.com
Mon Apr 30 15:40:46 UTC 2001


>>>>> "Federico" == Federico Bockel <fbockel at arnet.net.ar> writes:

    Federico> Im having problems when named refresh the zones, the
    Federico> service stop responding for some minutes, i have 3000+
    Federico> zones.
    Federico> Is it common that when slaves are refreshing
    Federico> zones they stop respondig for some minutes?(master
    Federico> continues responding ok)

It's not common. In fact it doesn't happen. A BIND[48] server will
stop answering queries while it loads a zone. In most circumstances,
this should be unnoticeable. It can be an issue when there are huge
numbers of resource records -- millions - or tens of thousands of
zones to load.

    Federico> What can i do to solve this?

Start by showing us what's in the logs. Have you any packet traces or
debugging logs? What makes you think that the name server has stopped
processing queries because of refresh checks? Are you sure there are
no other problems, like a routing error or a broken network?

    Federico> Any idea if the transfers-in and transfers-out will help?

You've not identified the actual problem, so it's hard to see how
tinkering with the number of zone transfers will help. Or any other
BIND option for that matter. I suppose you could try all possible
options in a process of trial and error and see which, if any, make a
difference. But you'd be better off doing a root cause analysis,
finding the underlying problem and fixing it. So far, it's not even
clear why the servers are not responding for a few minutes.


More information about the bind-users mailing list