Le 2014-01-20 09:00, Gert Doering a écrit :
I've run into this sort of problems a few years ago, but I used a
different solution: I didn't use mapped addresses but two separate
sockets, one for IPv4 and another for IPv6.
This *is* a long-term goal, though, to enable OpenVPN to listen on multiple
sockets in parallel (including "udp and tcp" in the same server), but this
part of the code is *old*, has way too many options, and runs on way too
many systems with their own specific surprises.
<rant>
In my experience, this is the single most difficult thing about
migrating a code base to dual-stack. Code bases that already deal with
multiple IPv4-only sockets are *much* easier to migrate to dual-stack
than those that expect only a single socket. An example of this is
Asterisk, which used only a single socket (I'm talking about SIP here).
We considered the alternatives, and eventually decided not to change the
whole architecture and instead use IPv4-mapped addresses on a single
socket to support dual-stack. And it sucked, of course. But at least it
was feasible (with a few tens of thousands line diff). Eventually the
whole SIP stuff was ripped out of Asterisk and replaced with a
third-party SIP stack that correctly dealt with multiple sockets and
thus didn't need IPv4-mappeed addresses to support IPv6.
</rant>
Simon
--
DTN made easy, lean, and smart --> http://postellation.viagenie.ca
NAT64/DNS64 open-source --> http://ecdysis.viagenie.ca
STUN/TURN server --> http://numb.viagenie.ca
