Re: Combining 4 C-class networks: how?
On Mon, 10 Jul 2000, Jameson Burt wrote:
> I run Debian potato Linux (upgrading it weekly) within a network
> that others have setup for Microsoft Windows 95/98.
> This network has (over the same ethernet cables)
> computers and printers covering 4 C-class networks
> as follows [first 2 digits altered],
> 10.129.206.0
> 10.129.207.0
> 10.129.208.0
> 10.129.209.0
> My computer has the IP address,
> 10.129.208.230
> HOW DO I GET MY LINUX COMPUTER TO WORK WELL IN A NETWORK OF 4 C-CLASSES?
>
First:
The 10.x.x.x network range is, due to standard ip ranges, class A.
You have used a subnet mask to divide it into 2^16 sub-ip-ranges, using
four of them.
So a simple solution might be (i've not actually done this) to treat it
like the class A network it actually is.
This requires reconfiguration of the windows machines however. (Or
multiple broadcast adresses, if possible.)
The free adress ranges for class C networks are 192.168.y.x , y=0..255.
(256 class C networks.)
> 2. Aliasing
> I largely followed ideas in the IP-alias mini-HOWTO.
>
> I turned on IP Aliasing in the kernel, then put the following in
> /etc/network/interfaces
>
> iface eth0 inet static
> address 10.129.208.230
> netmask 255.255.255.0
> network 10.129.208.0
> broadcast 10.129.208.255
> gateway 10.129.208.1
>
> iface eth0:0 inet static
> address 10.129.206.1 #but this is another computer's address
> netmask 255.255.255.0
> network 10.129.206.0
> broadcast 10.129.206.255
> gateway 10.129.208.1
>
> iface eth0:1 inet static
> address 10.129.207.1 #but this is another computer's address
> netmask 255.255.255.0
> network 10.129.207.0
> broadcast 10.129.207.255
> gateway 10.129.208.1
>
> iface eth0:2 inet static
> address 10.129.209.1 #but this is another computer's address
> netmask 255.255.255.0
> network 10.129.209.0
> broadcast 10.129.209.255
> gateway 10.129.208.1
>
I still believe IP aliasing should be done with eth0:0, eth0:1, eth0:2,
eth0:3 instead of eth0, eth0:0, eth0:1, eth0:2.
The documentation says nothing about the latter, as far as I know.
The non-IP-aliasing method of eth0 and the IP-aliasing method of eth0:0
may be mutually exclusive, or eth0 may be an alias for eth0:0 when kernel
IP aliasing is turned on.
> This approach worked well for Samba for 5 minutes until, I presume,
> other computers like 10.129.206.1 conflicted with
> my multiple assignment of IP addresses to the same ethernet card,
> including this very "address 10.129.206.1" above.
> Of course, I have but one IP address, so my eth0 card
> should be assigned but one IP address.
> I last tried this approach 2 days ago.
>
>
>
> So, you can see I would like all 4 C-class subnets to be somehow
> recognized as one network by my Debian computer,
> which would probably require the following settings,
>
> a. address of my eth0 interface
> 10.129.208.230
>
> I can get but one address for my computer from my administrators.
> I don't need all the IP aliasing that lets my computer respond
> to several IP address; I only need my computer to interract
> in a network that has 4 C-class networks.
>
This is in order to reduce netbios socializing, I guess. ("Hello! Here I
am! Does anybody hear me?". Does a huge amount of traffic on large
networks). Netbios isn't meant for larger networks and
wondiws doesn't support the tcp/ip-tools (for example) out of the box.
They should probably have used a WINS server instead. It centralizes the
netbios socializing and keeps it down.
Making clients have to listen to several ip ranges is downright idiotic
if you ask me, but you are in a "windows network" so who should expect
any less. (Apply the old joke: "How many microsoft engineers do you
need to change a light bulb? None. They just define darkness as a new
standard.")
> b. broadcast addresses
> 10.129.206.255
> 10.129.207.255
> 10.129.208.255
> 10.129.209.255
>
> I believe I need all these standard C-class broadcast addresses
> because all other computers on the network use one
> of these four lines as a broadcast address.
>
> c. netmask and network
> I don't know what I need for these.
> Indeed, with the problems I have with trying to solve
> my networking problem through IP subnetworking and IP aliasing,
> I am unsure what approach to use with the above (a) and (b) restrictions.
> Any ideas?
>
Harald
Philosophy, law, medicine and even theology, alas! I studied everything with
an ardent will and here I am, poor fool, just as far behind as ever.
No more advanced than before.
Goethe - Faust
Reply to: