It depends on what you need. Brokers can now be configured with external (firewall) address and advertise those in the ObjRefs they generate, we also support bi-directional connections (see the 7.5 Release Notes) which will allow a "server" to send requests over connections previously established by clients.
The remaining problem with tunneling is that the "client" needs to know enough to talk to the local end of the tunnel, instead of the address advertised by the "server". If you have just a handful of remote references that you could create by hand, you can just create them with the right addresses and all should work fine. But if you'll have new references created on the fly and passed back and forth, you really need some sort of client-side address translation, which would check the address in the ObjRef, whenever it is about to send a remote request to it, and instead of looking for a connection to the advertised address route it through the one going to the local end of the tunnel. It shouldn't be very difficult to extend the broker that way, but of course you'll need to get your hands dirty with the internals. It might even be possible to achieve by exploiting the numerous hooks that the broker provides (see broker events).
We have an AR for this, 49280: "[stst] add support for tunneling of STST communication, e.g. using it over stunnel", unfortunately no progress on that one still :(.
HTH,
Martin
C. David Shaffer wrote:
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc
Locked
