Kastus wrote:
Running tinydns on a loopback address for an internal domain is not strange at all. Then you can run dnscache on eth interface and point it to tinydns on loopback for that internal domain.
This allows to use the same host for multiple internal domains. For example, you have dev, QA, and prod environments. All use the same domain but should resolve to different IP addresses in different environments. Using multiple loopback addresses on the same host you can run tinydns for all your environments on the same host.
Fair enough. I've never used tinydns. Those times that I've had internal and external dns's, a firewall was involved, so they weren't on the same machine.
On top of that, run multiple instances of dnscache bound to different secondary addresses on eth interface. Then point DNS resolver to a diiferent dnscache depending on the environment (dev, qa, prod) and you get a way of testing your application in different environments using the same FQDNs.
I was always taught to stay away from loopback stuff, and just use alias eth interfaces as you mention here. I suppose in the grand scheme of things, one method is as good as another.