Dear Joris, >>BTW: I hate "solutions by counter questions". Therefore, it should not be a >>solution but just a question of interest >>why the physical reach-through of interfaces is your preferred solution. > >Because my hosts are virtuals, and it's very easy to me to provide them 3 >interfaces, mounting it by type=phys, and nothing to >do more to communicate with my containers (no routing, port forwarding...)
Does this mean that your "physical" interfaces are no real ones but emulated hardware by an outside virtualization platform? If this is the case, it may be in the responsibility of this layer when the traffic stops on this NICs. >Is it such a bad idea ? Another approach would be to use a bridge on the lxc host and a veth for each container and apply the same IP's to them. Because bridging is layer2 wiring, you also don't need any routing, port forwarding and so on. Without any background one may say that this approach would be the generic one. Of course, it might be more suitable for your environment if each Container have a manageable interface which is provided by the virtualization platform -- e.g. to use traffic accounting tools provided there. Greetings Guido _______________________________________________ lxc-users mailing list [email protected] http://lists.linuxcontainers.org/listinfo/lxc-users
