Hacker News new | past | comments | ask | show | jobs | submit login

If .corp is successfully resolving this change should have no impact, since the problem is only when .corp fails to resolve and the resolver falls back to .corp.com.



Sorry - you’re right. I misread.

But I still think failing over from corp to corp.com is a bug, not a feature.


> But I still think failing over from corp to corp.com is a bug, not a feature.

I think it was a mistake, but you probably can't fix it now. Too many random pieces of code are going to be relying on it.


That doesn't make sense. When .corp fails to resolve now, .corp.com is also a deadend because O'Connor turned the domain off, so nothing should happen to any software by not trying to access .corp.com.

Surely a rule can be made in the resolver code that specifically in the case of .corp failing to resolve, short-circuit it there and do not failover to internet domains.


Right, what you're suggesting is one implementation of what I meant by "the moral equivalent of adding these lines to /etc/hosts".

I think TimTheTinker was suggesting the much more general fix of breaking the fallback for all domain names, not just .corp domain names.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: