Stalling slave transfers
Cathy Almond
cathya at isc.org
Fri May 17 14:54:22 UTC 2013
On 15/05/13 15:58, Tony Finch wrote:
> Tom Sommer <mail at tomsommer.dk> wrote:
>>
>> That works fine, but I think I figured out the problem, it was due to
>> the server having acquired a 2nd (autodiscovered) IPv6 address, and it
>> was using that as transfer source. It would be very helpful if the
>> logfile said the actual source IP, and not just 0.0.0.0#53 or ::#0. That
>> would help debugging a lot.
>
> I have found that if you have multiple master addresses listed for a slave
> zone, named will not fall back to trying later addresses if the first one
> fails.
>
> Tony.
>
The speed of fall-back through the masters list may depend on whether or
not you set "try-tcp-refresh no;" in named.conf.
Another contributing factor is whether the failure mode is immediate
(ICMP error or connection failure) or has to time out from named's
perspective.
More information about the bind-users
mailing list