Hello Gert On Wed, 29 May 2002 16:38:37 +0200, Gert Doering <gert@space.net> said: F
Hi, On Wed, May 29, 2002 at 04:31:48PM +0200, Alexander Gall wrote:
I heave heard a number of "I think, I don't think, I suppose" statements on this, but no hard facts from the RFCs or code from common resolver libraries.
Ok, I forgot about wildcards <shudder>. With a wildcard, you can actually have a server return such a CNAME. However, whatever you specify as the target of the CNAME is static. To "redirect" an entire branch of the DNS tree, you need to modify the original query before you restart it. That's why DNAME had to change the algorithms of RFC1034, sections 4.3.2 and 5.3.3.
Am I missing something?
You could use one of the non-BIND servers that can actually generate arbitrary records on-the-fly. This could be used for the "proxy" solution.
Of course.
I'd still like to hear about hard facts / experiences why a CNAME in the middle of the path won't work (having seen that it does work for forward resolving - looks ugly, but people use it)
What are you referring to? Also, "forward" and "inverse" zones are not inherently different. -- Alex