Well, I'm not convinced:
the testbed looks like this:
client --> router --> director
the client generates random requests (192.168.7.0/24 as source-addresses).
The destination address is a VIP (192.168.11.5). The router forwards the packet
to the director.
The router / director are directly connected in a little subnet (192.168.8.0/24;
the router has IP-address 192.168.8.1 and
the director 192.168.8.2). Packets with destination-address the VIP are
to the director, but as the source address hasn't the same subnet as the the
incoming interface of the director
(192.168.8.0 subnet vs 192.168.7.0) the director did not forward the packets. I
had to change the subnet of the incoming interface to 192.168.0.0/16. I really
don't understand why this is a requirement of networking, because the
source-address doesn't matter in this case as packets are only forwarded until
the realserver is reached. That's why they invented routing.
Joseph Mack NA3T wrote:
> On Wed, 20 Jul 2005, Davy De Winter wrote:
>> thanks for the replies; finally I managed to get things work. I had to change
>> - the source-address of the clients must fall in the subnet-range of the
incoming interface at the director.
> this is one of the requirements for networking.
--- Dit bericht werd verzonden met de Webmail van Hogeschool Gent