I use mangle rules to set routing marks on PPTP interfaces so that different PPTP clients are routed using different tables. I guess people are trying to do something different from that but I’m not clear what?
It might help to get the feature addressed if somebody can clarify?
I’m not sure if you found some magic way to handle this or if you aren’t seeing our need.
basically this:
Scenario1:
I have 2 WAN interfaces. I want to connect via an SSTP tunnel from WAN1 to remote site A and from WAN2 to remote site B. I can’t actually tell the SSTP tunnel to use WAN1 or WAN2 as the source for the tunnel, instead I have to create a route to site B via WAN2’s gateway. This is possible but a hassle.
Scenario2:
I have 2 WAN interfaces. I want to connect via an SSTP tunnel from WAN1 to remote site A and from WAN2 to remote site A also. Now I cannot cheat by setting a route to site B through WAN2’s subnet. I cannot have these tunnels connect this way. This is a useful scenario when a site can only get 1.5Mb DSL lines and I want to bond two or three of these interfaces to the remote site. The remote site, the hub of my VPN cloud, has a 100Mb FDX connection.