The Linux NET-3-HOWTO states at section 5.1.4 that it is legal to use the same IP address on each interface of a single machine in some circumstances. I don't understand how that could be. Here it is what I guess to happen: 1) If two interfaces are attached to the same physical network they would compete for the same connections from other hosts: Both interfaces would answer the ARP requests with a different MAC and, correct me if I am making mistake, the host initiating the connection would use the MAC of the last ARP response, making impractical the other interface. In the case that the host having two interfaces starts the communication it would send its datagrams to the network and interface that first matches the destination address in the routing table. Again, the rest of the interfaces have no practical use. 2) If each of two interfaces are attached to different physical networks among which the host would function as a gateway, the gateway wouldn't be able to forward datagrams correctly. No matter what netmask is used for the gateway interfaces the network portion of the destination IP address would always match both routes in the routing table, or at least it would forward datagrams towards the interface that first matches the destination address in the routing table, since a lookup in the routing table only matches one entry. If two physical interfaces are used for the purpouse of doubling bandwidth, they should be used under an only OS network interface or with two different IP addresses. I suspect the NET-3-HOWTO assesment is in error. If it's not I would appreciate the explanation and some example. I'm developing user networking documentation, so this will help other people to understand the internals of Linux networking and what they can and cannot do with the current TCP/IP implementation. TIA. Zacarías - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
