On Monday 20 December 2010 09:37:48 Dale wrote: > I set it up like this. The modem uses DHCP to get the IP from AT&T. > My local IP from the modem is 192.168.1.2. Then the router has the > IP 192.168.2.1 for my connection to the puter. The IP of my puter > is 192.168.2.5. The next puter will be 192.168.2.6 or something > different anyway.
The one thing you didn't mention there is the outer address of your router. It needs to be 192.168.1.x where x is anything other than 2. It needs to be on the same network segment as the inner side of your modem. > I need to read up on the netmask thing some more. It's still murky > for sure. (What follows has grown rather long. I hope it doesn't come over too much as a lecture.) It's fairly straightforward once you get the hang of it. The address of a device is a 64-bit number, expressed as four 16-bit numbers joined with dots. It's just easier to read when split into chunks, but it is really a 64-bit number. As in decimal arithmetic, the right-hand digit is the least significant. An interface address consists of two parts: the leftmost part defines a group of addresses (the network part) and the rightmost part specifies the number of the interface in that group (the host part). The function of the network mask is to specify where the boundary is between the network part and the host part. Two conventions are used for expressing where that boundary is: the older method is to write, say, 255.255.255.0, which indicates that the first 24 bits (three eight-bit numbers - 255 is all-ones in eight bits) belong to the network and anything to the right of those can be allocated to interfaces in that network. That convention dates from the era of plenty of IP addresses in the world and goes along with Class A, B, C or D. A class A network has a mask of 255.0.0.0, class B has 255.255.0.0, class C has 255.255.255.0 and a class D (never used in the wild as far as I know) would have 255.255.255.255. Since the meteoric growth of the Internet this class scheme has become a handicap, and a finer division of network scope has become necessary, to allow use of, say, 255.255.255.248 as a net mask. Rather than specifying a plethora of new classes (we'd need anything up to 60), a shorthand notation has been invented in which we just append a number to an address to specify the number of bits that identify the network, with the rest identifying the host on it (strictly speaking, a host's interface on the network, as a host may have more than one interface - sometimes even on the same network). This scheme is known as CIDR notation. Thus your modem's inner address is, I assume, 192.168.1.2/24, which is the same as writing 192.168.1.2 with a mask of 255.255.255.0. That mask 255.255.255.248 I mentioned specifies 29 bits for the network address and three for the hosts on it; that's enough for six computers once the ..0 and ..7 addresses are reserved for network address and broadcast address. A lot of ISPs use such a scheme for allocating address ranges to their customers. > How's it look? Think it will work for a while? Once you've set your router's outer address correctly, yes. Sorry I was asleep overnight and had to leave you to the tender mercies of your compatriots. :-) Again, apologies if I've seemed to want to teach my grandmother to suck eggs. -- Rgds Peter. Linux Counter 5290, 1994-04-23.
