Questions

[Benoit Steiner]

1/ Very good to test the robustness of the vlanserver since a user will
sooner or later misconfigure its vlc ;) And also very good to train the
diffs to diagnose quickly a problem ;) 
2/ Why not ? Seems robust and reliable to me.
3/ Means that the vlanserver depends on the vlanbridge to be able to
work. Until now, there has been no such strong dependency.
4/ Even worse than 1, because the database is full of mistakes. 

About question 1, who cares anyway :) Seriously a user should only make
requests for its IP and then the server determines if there are other
machine to move. 

About question 2, the only solution that works is to let the vlanbridge
do the reverse ARP. But I don't  like it. It's up to you.

The biggest problem here is that when a computer boots the vlanbridge
assumes it's on vlan2. So what happens if this computer is connected to
a hub that is on vlan4 ? There are a lot of problems like this. We
should consider that a port on a switch is used by a group of machines
on the same vlan, not a single one. But then we have no easy way to
automaticly and dynamicly populate such groups. As a result, supporting
hubs is a real pain. We should really require everybody to buy a switch:
They're no more that expensive ;)

Benny


Damien Lucas wrote:
> 
> I was working on bridging all the macs known on a port
> (ie when you got a hub, you 'd like all you're computers to be bridged
>  when you ask for vlanchanging).
> 
> There is no problem to know all the macs (see telnet/info)
> BUT we need the IPs.
> 
> 4 solutions:
> 
> 1- The client sends the IP list
> 2- We make a reverse ARP
> 3- We ask to the VLANbridge
> 4- We ask to Webbase
> 
> These are the only 4 solutions I've seen. Do you have another one ?
> 
> The problems are:
> 1- What about the security (I can ask for bridging anyone ...) ?
> 2- What if the macs are not in the same VLAN as VLANserver ?
> 
> What do you think of that ?
> 
> --
> nitrox