Forwarders timeout.

[Kevin Darcy]

BIND 8.2.3 supposedly addresses this by selecting forwarders according to
RTT (round-trip-time), just like it picks nameservers according to
RTT when resolving iteratively. Presumably, this same logic was carried
forward to BIND 9 as well. I haven't personally verified whether this
works or not, I'm just going by the documentation...


- Kevin

johnliao at yahoo.com wrote:

> I have the following set up.
>
> NS1-internal(pri)- looks up only internal names, 10.10.10.10
> NS2-internal(sec)- looks up only internal names, 10.10.10.20
>
> NS1-external(pri)- looks up external names, 10.10.10.30
> NS2-external(sec)- looks up external names, 10.10.10.40
>
> I have have set the internal name servers to use the forwarding
> options.
>
> Something like this
>
>    options { forwarders {10.10.10.30; 10.10.10.40;};
>              forward only;
>    };
>
> To test the redundancy of the external name server. I kill the named
> process on NS1-external. I do a nslookup using one of the internal name
> servers. I eventually get a resolution back, but it take a long long
> time. Probably close to 2 minutes.
>
> My question is, Is there a way to decease the amount of time such that
> the request would move on to the next forwarder on the list if the
> first one is not available?
>
> So if I were to do an nslookup on www.yahoo.com. It'll search my
> internal name sever for the cache, if not found it'll forward the
> request it to my external name server. If my first external name server
> 10.10.10.30, is not available, it should  move onto the next name
> server 10.10.10.40 almost instantaneously.
>
> Any suggestions or other redundancy suggestions.