DNS resolution based on source network

[Matus UHLAR - fantomas]

On 27.09.10 19:38, Kevin Darcy wrote:
> Under certain limited circumstances, it might make more sense to put  
> both/all addresses under the same name, and then use the "sortlist"  
> mechanism to present those addresses in an order which is suitable for  
> particular clients.

certain? I'd say under most. It's always better to get rrset soertd in
network topological order, but when any of servers fails, it's good to have
backup.

If all servers are reachable, simple sortlist statement will be enough.
If they are not, you need different zones in different views.

> Among other things, this requires that all resolver/nameserver configs  
> be configured with the same sortlist configs, that there is no local  
> randomization or re-sorting of the address list,

I've had such problem some time ago (addresses were re-sorted in numeric
order), the suspect was libc or nss_lwres.

> that there are no negative consequences for the client or the client
> software to connect to the "wrong" address if the preferred one happens to
> be unavailable.

if there are negative cinsequencies of something like that, you/we need load
balancing, failover switching etc.


-- 
Matus UHLAR - fantomas, uhlar at fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
M$ Win's are shit, do not use it !