[dns-operations] Client retry behavour?
gall at switch.ch
Thu May 10 09:26:43 UTC 2007
On Thu, 10 May 2007 03:38:42 +1000 (EST), Bruce Campbell <bc-dns at vicious.dropbear.id.au> said:
> On Wed, 9 May 2007, Stephane Bortzmeyer wrote:
>> My experience on Unix is that the stub resolver included in the libc
>> takes a long time (several seconds, which is typically unbearable) to
>> timeout and to move to the second name server listed in
>> /etc/resolv.conf. Worse, since it has no memory, the problem appears
>> each time you run a new program.
> I've sometimes wondered why resolv.conf doesn't have an option to
> randomise the initial query order of nameservers. But then down that path
> lies the feeping creaturism of 'primary_nameserver' and its ilk.
As Paul has mentioned, there is the RES_ROTATE option. We use
options timeout:1 rotate
and two nameservers with anycast (currently two nodes for each one).
This has proven to be very stable and I haven't heard complaints about
delays during lookups from our users in years.
More information about the dns-operations