This little project kept me up late last night, and also accounted for a brief interruption in my server availability between 21h00 and 22h00.
I have set up a Teredo gateway, which gives limited IPv6 availability by way of an IPv4 tunnel, and which presents the local computer with a virtual IPv6 NIC. The ultimate goal of this exercise was, that people in the WWW who have native IPv6 access, should be able to access my site using this IPv6 address.
Right now the global address in question is ‘2001:0:53aa:64c:302a:2849:b9e5:30c7′, but it may change at any time.
While it seems superficially that I may have succeeded at achieving this goal, I am unable to give it a thorough test, because the Teredo gateway also does not allow loop-back access. For me to try loop-back access, I need to use the IPv6 address of the loop-back interface provided by my kernel, which is ‘::1′. The problem with this is, that it does not provide a thorough test, or even an adequate test. Certain single Web-pages display fine.
But in order for my blog, for example, to display correctly, it is not only the main, home page of the blog which must get served up, but also the many individual URLs embedded within that main page-view. And then what seems to happen with my own PC, is that even if I specified the IPv6 address literally as a URL to fetch (which requires embedding it in square brackets), all the embedded URLs are still resolved to the IPv4 DNS addresses by my browser, resulting in an inconsistency, which I believe will prevent any complex page from displaying.
The only way this blog will display correctly using its IPv6 address, is from a client, on which every reference to my host URL has an IPv6 resolution.